diff --git a/.claude/blackboard.md b/.claude/blackboard.md
index eb3aded7..1bcff2e2 100644
--- a/.claude/blackboard.md
+++ b/.claude/blackboard.md
@@ -1,3 +1,18 @@
+## 2026-09-24 — per-CPU SIMD inventory generated from LLVM's TableGen source
+
+`tools/gen_llvm_inventory.py` reads llvm-project's `X86.td`, `AArch64Processors.td`, `AArch64Features.td` (plus the instruction/predicate `.td` files) at the `llvmorg-*` tag matching `rustc -vV` (22.1.8 under the pinned 1.98.1), resolves every processor's features, and writes `tools/llvm-inventory/inventory.json` + `.claude/knowledge/llvm-cpu-inventory.md`. Same pattern as `gen_ternlog_bodies.py`: committed output, generator as provenance; default mode exits 1 if the output is stale.
+
+**Verified, not asserted:** `--verify-rustc` compares EVERY CPU (92 x86, 99 aarch64) with rustc's own `--print cfg -C target-cpu=...` over all features stable rustc exposes (56 / 37), using rustc's own Rust↔LLVM name table parsed from `rustc_codegen_llvm/src/llvm_util.rs` at the pinned release. 191/191 agree. The comparator failed first and drove five real fixes: `!listremove`; `Architecture64` implied list is the first list arg, not the last; aarch64 arch DefaultExts are part of rustc's cfg (apple-m4 `ssbs`); x86 ABI baseline `sse/sse2` is target-level; **tune records imply ISA features** (`TuneOryon` → FullFP16/FP16FML/SPE).
+
+**Findings the table makes visible (for crypto dispatch):**
+- `x86-64-v3` (our portable baseline) has no `aes` and no `pclmul` — compile-time dispatch cannot reach AES-NI or PCLMULQDQ on the v3 build.
+- `haswell` in LLVM has `pclmul` but not `aes`; `arrowlake` has `avxifma` (VEX IFMA) without AVX-512.
+- aarch64 `neoverse-v2` (Graviton4/Grace) and `cortex-x925` have no `aes` → no hardware AES and no 64-bit PMULL without an explicit `+aes`.
+
+**Primitive rows** (mul_lo32, var shift/permute u32, IFMA, clmul, AES round) name the LLVM instruction def and gating predicate; the generator ASSERTS they still exist in the `.td` files, so an LLVM rename fails loudly.
+
+**Loose ends:** the `backend` column mirrors the `simd.rs` cfg ladder by hand (not parsed); no wasm rows (wasm has no processor table); no CI job runs the staleness check yet; rustc builds on rust-lang's LLVM fork (`rustc/22.1-2026-05-19`) — its AArch64 `.td` files matched the upstream tag byte-for-byte here, and `--verify-rustc` is the guard if that ever diverges.
+
 ## 2026-09-23 (21) — ternlog → Count/Any without a mask: a slice loop, NOT a new ISA primitive
 
 Question asked (lance-graph #1270 prompt): what is the smallest T1 operation that lets an arbitrary 2/3-input Boolean membership END in Count/Any without writing a mask — and does existing `U64x8` composition already do it register-only?
diff --git a/.claude/knowledge/llvm-cpu-inventory.md b/.claude/knowledge/llvm-cpu-inventory.md
new file mode 100644
index 00000000..9b85cfc0
--- /dev/null
+++ b/.claude/knowledge/llvm-cpu-inventory.md
@@ -0,0 +1,211 @@
+# Per-CPU SIMD inventory — generated from LLVM's TableGen source
+
+> **GENERATED** by `tools/gen_llvm_inventory.py` from llvm-project `llvmorg-22.1.8`
+> (the tag matching `rustc -vV` under this repo's pinned toolchain). Do not edit by hand:
+> run `python3 tools/gen_llvm_inventory.py --write`. `--verify-rustc` checks every row
+> against rustc's own `--print cfg -C target-cpu=...`.
+
+`backend` mirrors the `simd.rs` cfg ladder. Primitive columns list the instruction(s)
+that implement the primitive on that CPU; `—` means the CPU has none and a portable
+fallback is required.
+
+## x86 (92 CPUs)
+
+| cpu | backend | avx2 | avx512f | avx512vl | avx512ifma | avxifma | pclmul | vpclmulqdq | aes | vaes | gfni | mul_lo32 | shift_var_u32 | permute_var_u32 | ifma52 | clmul | aes_round |
+|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
+| `alderlake` | simd_avx2 | ✓ |  |  |  |  | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | — | PCLMULQDQ | AESENC |
+| `arrowlake` | simd_avx2 | ✓ |  |  |  | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (VEX) | PCLMULQDQ | AESENC |
+| `arrowlake-s` | simd_avx2 | ✓ |  |  |  | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (VEX) | PCLMULQDQ | AESENC |
+| `arrowlake_s` | simd_avx2 | ✓ |  |  |  | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (VEX) | PCLMULQDQ | AESENC |
+| `atom` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  |  |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | — | — |
+| `atom_sse4_2` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  | ✓ |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | PCLMULQDQ | — |
+| `atom_sse4_2_movbe` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  | ✓ |  | ✓ |  |  | PMULUDQ (xmm, SSE2) | — | — | — | PCLMULQDQ | AESENC |
+| `bdver1` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  | ✓ |  | ✓ |  |  | PMULUDQ (xmm, SSE2) | — | — | — | PCLMULQDQ | AESENC |
+| `bdver2` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  | ✓ |  | ✓ |  |  | PMULUDQ (xmm, SSE2) | — | — | — | PCLMULQDQ | AESENC |
+| `bdver3` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  | ✓ |  | ✓ |  |  | PMULUDQ (xmm, SSE2) | — | — | — | PCLMULQDQ | AESENC |
+| `bdver4` | simd_avx2 | ✓ |  |  |  |  | ✓ |  | ✓ |  |  | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | — | PCLMULQDQ | AESENC |
+| `bonnell` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  |  |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | — | — |
+| `broadwell` | simd_avx2 | ✓ |  |  |  |  | ✓ |  |  |  |  | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | — | PCLMULQDQ | — |
+| `btver2` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  | ✓ |  | ✓ |  |  | PMULUDQ (xmm, SSE2) | — | — | — | PCLMULQDQ | AESENC |
+| `cannonlake` | simd_avx512 | ✓ | ✓ | ✓ | ✓ |  | ✓ |  | ✓ |  |  | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (EVEX) | PCLMULQDQ | AESENC |
+| `cascadelake` | simd_avx512 | ✓ | ✓ | ✓ |  |  | ✓ |  | ✓ |  |  | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | — | PCLMULQDQ | AESENC |
+| `clearwaterforest` | simd_avx2 | ✓ |  |  |  | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (VEX) | PCLMULQDQ | AESENC |
+| `cooperlake` | simd_avx512 | ✓ | ✓ | ✓ |  |  | ✓ |  | ✓ |  |  | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | — | PCLMULQDQ | AESENC |
+| `core-avx-i` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  | ✓ |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | PCLMULQDQ | — |
+| `core-avx2` | simd_avx2 | ✓ |  |  |  |  | ✓ |  |  |  |  | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | — | PCLMULQDQ | — |
+| `core2` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  |  |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | — | — |
+| `core_2_duo_sse4_1` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  |  |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | — | — |
+| `core_2_duo_ssse3` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  |  |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | — | — |
+| `core_2nd_gen_avx` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  | ✓ |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | PCLMULQDQ | — |
+| `core_3rd_gen_avx` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  | ✓ |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | PCLMULQDQ | — |
+| `core_4th_gen_avx` | simd_avx2 | ✓ |  |  |  |  | ✓ |  |  |  |  | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | — | PCLMULQDQ | — |
+| `core_4th_gen_avx_tsx` | simd_avx2 | ✓ |  |  |  |  | ✓ |  |  |  |  | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | — | PCLMULQDQ | — |
+| `core_5th_gen_avx` | simd_avx2 | ✓ |  |  |  |  | ✓ |  |  |  |  | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | — | PCLMULQDQ | — |
+| `core_5th_gen_avx_tsx` | simd_avx2 | ✓ |  |  |  |  | ✓ |  |  |  |  | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | — | PCLMULQDQ | — |
+| `core_aes_pclmulqdq` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  | ✓ |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | PCLMULQDQ | — |
+| `core_i7_sse4_2` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  |  |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | — | — |
+| `corei7` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  |  |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | — | — |
+| `corei7-avx` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  | ✓ |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | PCLMULQDQ | — |
+| `diamondrapids` | simd_avx512 | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (EVEX)<br>VPMADD52LUQ/HUQ (VEX) | PCLMULQDQ<br>VPCLMULQDQ (zmm) | AESENC<br>VAESENC (zmm) |
+| `emeraldrapids` | simd_avx512 | ✓ | ✓ | ✓ | ✓ |  | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (EVEX) | PCLMULQDQ<br>VPCLMULQDQ (zmm) | AESENC<br>VAESENC (zmm) |
+| `generic` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  |  |  |  |  |  | — | — | — | — | — | — |
+| `goldmont` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  | ✓ |  | ✓ |  |  | PMULUDQ (xmm, SSE2) | — | — | — | PCLMULQDQ | AESENC |
+| `goldmont-plus` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  | ✓ |  | ✓ |  |  | PMULUDQ (xmm, SSE2) | — | — | — | PCLMULQDQ | AESENC |
+| `goldmont_plus` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  | ✓ |  | ✓ |  |  | PMULUDQ (xmm, SSE2) | — | — | — | PCLMULQDQ | AESENC |
+| `gracemont` | simd_avx2 | ✓ |  |  |  |  | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | — | PCLMULQDQ | AESENC |
+| `grandridge` | simd_avx2 | ✓ |  |  |  | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (VEX) | PCLMULQDQ | AESENC |
+| `graniterapids` | simd_avx512 | ✓ | ✓ | ✓ | ✓ |  | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (EVEX) | PCLMULQDQ<br>VPCLMULQDQ (zmm) | AESENC<br>VAESENC (zmm) |
+| `graniterapids-d` | simd_avx512 | ✓ | ✓ | ✓ | ✓ |  | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (EVEX) | PCLMULQDQ<br>VPCLMULQDQ (zmm) | AESENC<br>VAESENC (zmm) |
+| `graniterapids_d` | simd_avx512 | ✓ | ✓ | ✓ | ✓ |  | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (EVEX) | PCLMULQDQ<br>VPCLMULQDQ (zmm) | AESENC<br>VAESENC (zmm) |
+| `haswell` | simd_avx2 | ✓ |  |  |  |  | ✓ |  |  |  |  | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | — | PCLMULQDQ | — |
+| `icelake-client` | simd_avx512 | ✓ | ✓ | ✓ | ✓ |  | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (EVEX) | PCLMULQDQ<br>VPCLMULQDQ (zmm) | AESENC<br>VAESENC (zmm) |
+| `icelake-server` | simd_avx512 | ✓ | ✓ | ✓ | ✓ |  | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (EVEX) | PCLMULQDQ<br>VPCLMULQDQ (zmm) | AESENC<br>VAESENC (zmm) |
+| `icelake_client` | simd_avx512 | ✓ | ✓ | ✓ | ✓ |  | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (EVEX) | PCLMULQDQ<br>VPCLMULQDQ (zmm) | AESENC<br>VAESENC (zmm) |
+| `icelake_server` | simd_avx512 | ✓ | ✓ | ✓ | ✓ |  | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (EVEX) | PCLMULQDQ<br>VPCLMULQDQ (zmm) | AESENC<br>VAESENC (zmm) |
+| `ivybridge` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  | ✓ |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | PCLMULQDQ | — |
+| `knl` | simd_avx512 | ✓ | ✓ |  |  |  | ✓ |  | ✓ |  |  | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | — | PCLMULQDQ | AESENC |
+| `knm` | simd_avx512 | ✓ | ✓ |  |  |  | ✓ |  | ✓ |  |  | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | — | PCLMULQDQ | AESENC |
+| `lunarlake` | simd_avx2 | ✓ |  |  |  | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (VEX) | PCLMULQDQ | AESENC |
+| `meteorlake` | simd_avx2 | ✓ |  |  |  |  | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | — | PCLMULQDQ | AESENC |
+| `mic_avx512` | simd_avx512 | ✓ | ✓ |  |  |  | ✓ |  | ✓ |  |  | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | — | PCLMULQDQ | AESENC |
+| `nehalem` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  |  |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | — | — |
+| `nocona` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  |  |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | — | — |
+| `novalake` | simd_avx512 | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (EVEX)<br>VPMADD52LUQ/HUQ (VEX) | PCLMULQDQ<br>VPCLMULQDQ (zmm) | AESENC<br>VAESENC (zmm) |
+| `pantherlake` | simd_avx2 | ✓ |  |  |  | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (VEX) | PCLMULQDQ | AESENC |
+| `penryn` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  |  |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | — | — |
+| `pentium-m` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  |  |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | — | — |
+| `pentium4` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  |  |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | — | — |
+| `pentium4m` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  |  |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | — | — |
+| `pentium_4` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  |  |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | — | — |
+| `pentium_4_sse3` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  |  |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | — | — |
+| `pentium_m` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  |  |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | — | — |
+| `prescott` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  |  |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | — | — |
+| `raptorlake` | simd_avx2 | ✓ |  |  |  |  | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | — | PCLMULQDQ | AESENC |
+| `rocketlake` | simd_avx512 | ✓ | ✓ | ✓ | ✓ |  | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (EVEX) | PCLMULQDQ<br>VPCLMULQDQ (zmm) | AESENC<br>VAESENC (zmm) |
+| `sandybridge` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  | ✓ |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | PCLMULQDQ | — |
+| `sapphirerapids` | simd_avx512 | ✓ | ✓ | ✓ | ✓ |  | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (EVEX) | PCLMULQDQ<br>VPCLMULQDQ (zmm) | AESENC<br>VAESENC (zmm) |
+| `sierraforest` | simd_avx2 | ✓ |  |  |  | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (VEX) | PCLMULQDQ | AESENC |
+| `silvermont` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  | ✓ |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | PCLMULQDQ | — |
+| `skx` | simd_avx512 | ✓ | ✓ | ✓ |  |  | ✓ |  | ✓ |  |  | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | — | PCLMULQDQ | AESENC |
+| `skylake` | simd_avx2 | ✓ |  |  |  |  | ✓ |  | ✓ |  |  | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | — | PCLMULQDQ | AESENC |
+| `skylake-avx512` | simd_avx512 | ✓ | ✓ | ✓ |  |  | ✓ |  | ✓ |  |  | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | — | PCLMULQDQ | AESENC |
+| `skylake_avx512` | simd_avx512 | ✓ | ✓ | ✓ |  |  | ✓ |  | ✓ |  |  | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | — | PCLMULQDQ | AESENC |
+| `slm` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  | ✓ |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | PCLMULQDQ | — |
+| `tigerlake` | simd_avx512 | ✓ | ✓ | ✓ | ✓ |  | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (EVEX) | PCLMULQDQ<br>VPCLMULQDQ (zmm) | AESENC<br>VAESENC (zmm) |
+| `tremont` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  | ✓ |  | ✓ |  | ✓ | PMULUDQ (xmm, SSE2) | — | — | — | PCLMULQDQ | AESENC |
+| `westmere` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  | ✓ |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | PCLMULQDQ | — |
+| `wildcatlake` | simd_avx2 | ✓ |  |  |  | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (VEX) | PCLMULQDQ | AESENC |
+| `x86-64` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  |  |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | — | — |
+| `x86-64-v2` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  |  |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | — | — |
+| `x86-64-v3` | simd_avx2 | ✓ |  |  |  |  |  |  |  |  |  | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | — | — | — |
+| `x86-64-v4` | simd_avx512 | ✓ | ✓ | ✓ |  |  |  |  |  |  |  | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | — | — | — |
+| `yonah` | simd_avx2 (no avx2: array polyfill) |  |  |  |  |  |  |  |  |  |  | PMULUDQ (xmm, SSE2) | — | — | — | — | — |
+| `znver1` | simd_avx2 | ✓ |  |  |  |  | ✓ |  | ✓ |  |  | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | — | PCLMULQDQ | AESENC |
+| `znver2` | simd_avx2 | ✓ |  |  |  |  | ✓ |  | ✓ |  |  | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | — | PCLMULQDQ | AESENC |
+| `znver3` | simd_avx2 | ✓ |  |  |  |  | ✓ | ✓ | ✓ | ✓ |  | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2) | VPSLLVD / VPSRLVD | VPERMD | — | PCLMULQDQ | AESENC |
+| `znver4` | simd_avx512 | ✓ | ✓ | ✓ | ✓ |  | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (EVEX) | PCLMULQDQ<br>VPCLMULQDQ (zmm) | AESENC<br>VAESENC (zmm) |
+| `znver5` | simd_avx512 | ✓ | ✓ | ✓ | ✓ |  | ✓ | ✓ | ✓ | ✓ | ✓ | PMULUDQ (xmm, SSE2)<br>VPMULUDQ (ymm, AVX2)<br>VPMULUDQ (zmm, AVX-512F) | VPSLLVD / VPSRLVD | VPERMD | VPMADD52LUQ/HUQ (EVEX) | PCLMULQDQ<br>VPCLMULQDQ (zmm) | AESENC<br>VAESENC (zmm) |
+
+## aarch64 (99 CPUs)
+
+| cpu | backend | neon | dotprod | i8mm | aes | sha2 | sve2 | mul_lo32 | shift_var_u32 | permute_var_u32 | ifma52 | clmul | aes_round |
+|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
+| `a64fx` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `ampere1` | simd_neon +dotprod | ✓ | ✓ | ✓ | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `ampere1a` | simd_neon +dotprod | ✓ | ✓ | ✓ | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `ampere1b` | simd_neon +dotprod | ✓ | ✓ | ✓ | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `ampere1c` | simd_neon +dotprod | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-a10` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-a11` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-a12` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-a13` | simd_neon +dotprod | ✓ | ✓ |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-a14` | simd_neon +dotprod | ✓ | ✓ |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-a15` | simd_neon +dotprod | ✓ | ✓ | ✓ | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-a16` | simd_neon +dotprod | ✓ | ✓ | ✓ | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-a17` | simd_neon +dotprod | ✓ | ✓ | ✓ | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-a18` | simd_neon +dotprod | ✓ | ✓ | ✓ | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-a19` | simd_neon +dotprod | ✓ | ✓ | ✓ | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-a7` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-a8` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-a9` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-latest` | simd_neon +dotprod | ✓ | ✓ | ✓ | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-m1` | simd_neon +dotprod | ✓ | ✓ |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-m2` | simd_neon +dotprod | ✓ | ✓ | ✓ | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-m3` | simd_neon +dotprod | ✓ | ✓ | ✓ | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-m4` | simd_neon +dotprod | ✓ | ✓ | ✓ | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-m5` | simd_neon +dotprod | ✓ | ✓ | ✓ | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-s10` | simd_neon +dotprod | ✓ | ✓ | ✓ | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-s4` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-s5` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-s6` | simd_neon +dotprod | ✓ | ✓ |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-s7` | simd_neon +dotprod | ✓ | ✓ |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-s8` | simd_neon +dotprod | ✓ | ✓ |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `apple-s9` | simd_neon +dotprod | ✓ | ✓ | ✓ | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `c1-nano` | simd_neon +dotprod | ✓ | ✓ | ✓ |  |  | ✓ | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `c1-premium` | simd_neon +dotprod | ✓ | ✓ | ✓ |  |  | ✓ | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `c1-pro` | simd_neon +dotprod | ✓ | ✓ | ✓ |  |  | ✓ | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `c1-ultra` | simd_neon +dotprod | ✓ | ✓ | ✓ |  |  | ✓ | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `carmel` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `cobalt-100` | simd_neon +dotprod | ✓ | ✓ | ✓ |  |  | ✓ | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `cortex-a320` | simd_neon +dotprod | ✓ | ✓ | ✓ |  |  | ✓ | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `cortex-a34` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `cortex-a35` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `cortex-a510` | simd_neon +dotprod | ✓ | ✓ | ✓ |  |  | ✓ | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `cortex-a520` | simd_neon +dotprod | ✓ | ✓ | ✓ |  |  | ✓ | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `cortex-a520ae` | simd_neon +dotprod | ✓ | ✓ | ✓ |  |  | ✓ | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `cortex-a53` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `cortex-a55` | simd_neon +dotprod | ✓ | ✓ |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `cortex-a57` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `cortex-a65` | simd_neon +dotprod | ✓ | ✓ |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `cortex-a65ae` | simd_neon +dotprod | ✓ | ✓ |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `cortex-a710` | simd_neon +dotprod | ✓ | ✓ | ✓ |  |  | ✓ | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `cortex-a715` | simd_neon +dotprod | ✓ | ✓ | ✓ |  |  | ✓ | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `cortex-a72` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `cortex-a720` | simd_neon +dotprod | ✓ | ✓ | ✓ |  |  | ✓ | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `cortex-a720ae` | simd_neon +dotprod | ✓ | ✓ | ✓ |  |  | ✓ | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `cortex-a725` | simd_neon +dotprod | ✓ | ✓ | ✓ |  |  | ✓ | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `cortex-a73` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `cortex-a75` | simd_neon +dotprod | ✓ | ✓ |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `cortex-a76` | simd_neon +dotprod | ✓ | ✓ |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `cortex-a76ae` | simd_neon +dotprod | ✓ | ✓ |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `cortex-a77` | simd_neon +dotprod | ✓ | ✓ |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `cortex-a78` | simd_neon +dotprod | ✓ | ✓ |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `cortex-a78ae` | simd_neon +dotprod | ✓ | ✓ |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `cortex-a78c` | simd_neon +dotprod | ✓ | ✓ |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `cortex-r82` | simd_neon +dotprod | ✓ | ✓ |  |  |  |  | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `cortex-r82ae` | simd_neon +dotprod | ✓ | ✓ |  |  |  |  | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `cortex-x1` | simd_neon +dotprod | ✓ | ✓ |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `cortex-x1c` | simd_neon +dotprod | ✓ | ✓ |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `cortex-x2` | simd_neon +dotprod | ✓ | ✓ | ✓ |  |  | ✓ | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `cortex-x3` | simd_neon +dotprod | ✓ | ✓ | ✓ |  |  | ✓ | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `cortex-x4` | simd_neon +dotprod | ✓ | ✓ | ✓ |  |  | ✓ | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `cortex-x925` | simd_neon +dotprod | ✓ | ✓ | ✓ |  |  | ✓ | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `cyclone` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `exynos-m3` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `exynos-m4` | simd_neon +dotprod | ✓ | ✓ |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `exynos-m5` | simd_neon +dotprod | ✓ | ✓ |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `falkor` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `fujitsu-monaka` | simd_neon +dotprod | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `gb10` | simd_neon +dotprod | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `generic` | simd_neon | ✓ |  |  |  |  |  | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `grace` | simd_neon +dotprod | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `kryo` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `neoverse-512tvb` | simd_neon +dotprod | ✓ | ✓ | ✓ | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `neoverse-e1` | simd_neon +dotprod | ✓ | ✓ |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `neoverse-n1` | simd_neon +dotprod | ✓ | ✓ |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `neoverse-n2` | simd_neon +dotprod | ✓ | ✓ | ✓ |  |  | ✓ | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `neoverse-n3` | simd_neon +dotprod | ✓ | ✓ | ✓ |  |  | ✓ | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `neoverse-v1` | simd_neon +dotprod | ✓ | ✓ | ✓ | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `neoverse-v2` | simd_neon +dotprod | ✓ | ✓ | ✓ |  |  | ✓ | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `neoverse-v3` | simd_neon +dotprod | ✓ | ✓ | ✓ |  |  | ✓ | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `neoverse-v3ae` | simd_neon +dotprod | ✓ | ✓ | ✓ |  |  | ✓ | UMULL / UMULL2 | USHL | TBL | — | — | — |
+| `olympus` | simd_neon +dotprod | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `oryon-1` | simd_neon +dotprod | ✓ | ✓ | ✓ | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `saphira` | simd_neon +dotprod | ✓ | ✓ |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `thunderx` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `thunderx2t99` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `thunderx3t110` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `thunderxt81` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `thunderxt83` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `thunderxt88` | simd_neon | ✓ |  |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
+| `tsv110` | simd_neon +dotprod | ✓ | ✓ |  | ✓ | ✓ |  | UMULL / UMULL2 | USHL | TBL | — | PMULL (64x64->128) | AESE + AESMC |
diff --git a/tools/gen_llvm_inventory.py b/tools/gen_llvm_inventory.py
new file mode 100644
index 00000000..92406c61
--- /dev/null
+++ b/tools/gen_llvm_inventory.py
@@ -0,0 +1,417 @@
+#!/usr/bin/env python3
+"""Generate the per-CPU SIMD feature inventory from LLVM's own TableGen source.
+
+WHY THIS EXISTS. ndarray dispatches at compile time on `cfg(target_feature)`.
+The set of features a given `-C target-cpu=X` turns on is not ours to decide:
+it is LLVM's processor table, compiled into rustc. That table is public source
+in llvm-project (`X86.td`, `AArch64Processors.td`, `AArch64Features.td`), so
+it can be READ instead of maintained by hand or re-derived by compiling probes.
+
+This script reads those `.td` files at the `llvmorg-*` tag that matches the
+LLVM rustc itself embeds (from `rustc -vV`, so `rust-toolchain.toml` pins it),
+resolves every processor's feature list (the `!listconcat` inheritance chains,
+the `foreach P = [...]` alias blocks, `ProcessorAlias`, and each feature's
+transitive implied features), and writes:
+
+  tools/llvm-inventory/inventory.json        full per-CPU feature sets
+  .claude/knowledge/llvm-cpu-inventory.md    the table a human reads
+
+For each CPU it also records which ndarray backend the `simd.rs` cfg ladder
+selects, and whether each SIMD primitive that the crypto references
+(curve25519-dalek, poly1305, argon2) need has its instruction on that CPU.
+
+The primitive table below names, per primitive, the LLVM instruction def and
+the LLVM predicate/feature that gates it. Those names are ASSERTED to exist in
+the fetched `.td` files, so an LLVM rename fails this script loudly instead of
+silently leaving a stale row.
+
+Like `gen_ternlog_bodies.py`, the output is committed and this script is its
+provenance.
+
+  python3 tools/gen_llvm_inventory.py --write          regenerate the outputs
+  python3 tools/gen_llvm_inventory.py                  check they are current (exit 1 if stale)
+  python3 tools/gen_llvm_inventory.py --verify-rustc   compare EVERY CPU against rustc's own
+                                                       `--print cfg -C target-cpu=...`
+
+Sources are fetched from raw.githubusercontent.com into
+`target/llvm-inventory/<tag>/` (gitignored); `--llvm-src DIR` reads a local
+llvm-project checkout instead.
+
+CAVEAT. rustc builds against rust-lang's LLVM fork, not the upstream tag. The
+fork carries patches, so the upstream tag is an approximation of what rustc
+embeds. `--verify-rustc` is the check that the approximation holds for the
+features rustc exposes.
+"""
+import argparse
+import json
+import pathlib
+import re
+import subprocess
+import sys
+import urllib.request
+
+ROOT = pathlib.Path(__file__).resolve().parent.parent
+OUT_JSON = ROOT / "tools" / "llvm-inventory" / "inventory.json"
+OUT_MD = ROOT / ".claude" / "knowledge" / "llvm-cpu-inventory.md"
+CACHE = ROOT / "target" / "llvm-inventory"
+
+TD = {
+    "x86": [
+        "llvm/lib/Target/X86/X86.td",
+        "llvm/lib/Target/X86/X86InstrAVX512.td",
+        "llvm/lib/Target/X86/X86InstrSSE.td",
+        "llvm/lib/Target/X86/X86InstrPredicates.td",
+    ],
+    "aarch64": [
+        "llvm/lib/Target/AArch64/AArch64Features.td",
+        "llvm/lib/Target/AArch64/AArch64Processors.td",
+        "llvm/lib/Target/AArch64/AArch64InstrInfo.td",
+    ],
+}
+RUST_TARGET = {"x86": "x86_64-unknown-linux-gnu", "aarch64": "aarch64-unknown-linux-gnu"}
+
+# Primitives the crypto references need, and what gates each one in LLVM.
+#   insn: (file basename, regex that must match the instruction's def)
+#   gate: (file basename, regex that must match the predicate/feature def)
+#   needs: LLVM feature names that must ALL be present on the CPU
+# Several rows per primitive = alternative encodings; a CPU has the primitive
+# if any row's `needs` is satisfied.
+PRIMITIVES = [
+    ("mul_lo32", "x86", "PMULUDQ (xmm, SSE2)", ("X86InstrSSE.td", r"defm PMULUDQ\b"), None, ["sse2"]),
+    ("mul_lo32", "x86", "VPMULUDQ (ymm, AVX2)", ("X86InstrSSE.td", r"defm PMULUDQ\b"), None, ["avx2"]),
+    ("mul_lo32", "x86", "VPMULUDQ (zmm, AVX-512F)", ("X86InstrAVX512.td", r"defm VPMULUDQ\b"), None, ["avx512f"]),
+    ("mul_lo32", "aarch64", "UMULL / UMULL2", ("AArch64InstrInfo.td", r"defm UMULL\s*:"), None, ["neon"]),
+    ("shift_var_u32", "x86", "VPSLLVD / VPSRLVD", ("X86InstrSSE.td", r"defm VPSLLVD\b"), None, ["avx2"]),
+    ("shift_var_u32", "aarch64", "USHL", ("AArch64InstrInfo.td", r"defm USHL\s*:"), None, ["neon"]),
+    ("permute_var_u32", "x86", "VPERMD", ("X86InstrSSE.td", r"defm VPERMD\b"), None, ["avx2"]),
+    ("permute_var_u32", "aarch64", "TBL", ("AArch64InstrInfo.td", r"defm TBL\s*:"), None, ["neon"]),
+    ("ifma52", "x86", "VPMADD52LUQ/HUQ (EVEX)", ("X86InstrAVX512.td", r"defm VPMADD52LUQ\b"),
+     ("X86InstrPredicates.td", r"def HasIFMA\b"), ["avx512ifma"]),
+    ("ifma52", "x86", "VPMADD52LUQ/HUQ (VEX)", ("X86InstrSSE.td", r"defm VPMADD52LUQ\b"),
+     ("X86InstrPredicates.td", r"def HasAVXIFMA\b"), ["avxifma"]),
+    ("clmul", "x86", "PCLMULQDQ", ("X86InstrSSE.td", r"PCLMULQDQ"),
+     ("X86InstrPredicates.td", r"def HasPCLMUL\b"), ["pclmul"]),
+    ("clmul", "x86", "VPCLMULQDQ (zmm)", ("X86InstrAVX512.td", r"defm VPCLMULQDQZ\b"),
+     ("X86InstrPredicates.td", r"def HasVPCLMULQDQ\b"), ["vpclmulqdq", "avx512f"]),
+    ("clmul", "aarch64", "PMULL (64x64->128)", ("AArch64InstrInfo.td", r"defm PMULL\s*:"),
+     ("AArch64Features.td", r'def FeatureAES\b[^;]*FEAT_PMULL'), ["aes"]),
+    ("aes_round", "x86", "AESENC", ("X86InstrSSE.td", r"defm VAESENC\b"),
+     ("X86InstrPredicates.td", r"def HasAES\b"), ["aes"]),
+    ("aes_round", "x86", "VAESENC (zmm)", ("X86InstrAVX512.td", r"defm VAESENC\b"),
+     ("X86InstrPredicates.td", r"def HasVAES\b"), ["vaes", "avx512f"]),
+    ("aes_round", "aarch64", "AESE + AESMC", ("AArch64InstrInfo.td", r"def AESErr\b"),
+     ("AArch64Features.td", r"def FeatureAES\b"), ["aes"]),
+]
+PRIM_ORDER = ["mul_lo32", "shift_var_u32", "permute_var_u32", "ifma52", "clmul", "aes_round"]
+
+# Features reported in the summary columns.
+SHOWN = {"x86": ["avx2", "avx512f", "avx512vl", "avx512ifma", "avxifma", "pclmul",
+                 "vpclmulqdq", "aes", "vaes", "gfni"],
+         "aarch64": ["neon", "dotprod", "i8mm", "aes", "sha2", "sve2"]}
+
+
+def rustc(*args):
+    return subprocess.run(["rustc", *args], cwd=ROOT, capture_output=True, text=True, check=True).stdout
+
+
+def llvm_tag():
+    m = re.search(r"LLVM version: (\d+\.\d+\.\d+)", rustc("-vV"))
+    if not m:
+        sys.exit("cannot read the LLVM version from `rustc -vV`")
+    return "llvmorg-" + m.group(1)
+
+
+def load(tag, src_dir):
+    out = {}
+    for arch, files in TD.items():
+        for rel in files:
+            name = pathlib.PurePosixPath(rel).name
+            if src_dir:
+                text = (pathlib.Path(src_dir) / rel).read_text()
+            else:
+                cached = CACHE / tag / name
+                if not cached.exists():
+                    cached.parent.mkdir(parents=True, exist_ok=True)
+                    url = f"https://raw.githubusercontent.com/llvm/llvm-project/{tag}/{rel}"
+                    with urllib.request.urlopen(url, timeout=60) as r:
+                        cached.write_bytes(r.read())
+                text = cached.read_text()
+            out[name] = text
+    return out
+
+
+def strip_comments(t):
+    return re.sub(r"//[^\n]*", "", t)
+
+
+def expand_foreach(t):
+    """`foreach P = ["a", "b"] in { def : ProcModel<P, ...>; }` -> one def per name."""
+    extra = []
+    for m in re.finditer(r"foreach\s+(\w+)\s*=\s*\[([^\]]*)\]\s*in\s*\{(.*?)\n\}", t, re.S):
+        var, body = m.group(1), m.group(3)
+        for name in re.findall(r'"([^"]+)"', m.group(2)):
+            extra.append(re.sub(r"<\s*" + var + r"\s*,", f'<"{name}",', body))
+    return t + "\n" + "\n".join(extra)
+
+
+def split_top(s):
+    parts, depth, cur = [], 0, ""
+    for ch in s:
+        if ch in "([<":
+            depth += 1
+        elif ch in ")]>":
+            depth -= 1
+        if ch == "," and depth == 0:
+            parts.append(cur)
+            cur = ""
+        else:
+            cur += ch
+    parts.append(cur)
+    return parts
+
+
+def resolve(text):
+    """Every processor -> set of LLVM feature names (implied features included)."""
+    t = expand_foreach(strip_comments(text))
+    features = {}
+    # Every feature-like def: the NAME is its first string argument, and the
+    # IMPLIED features are its first plain list argument. That holds for
+    # SubtargetFeature<name, attr, value, desc, [implied]>, Extension*<name,
+    # spelling, FEAT_, desc, [implied]> and Architecture64<major, minor,
+    # profile, name, [implied], !listconcat(defaults)>. Taking the LAST list
+    # instead reads Architecture64's -march defaults as implied features.
+    for m in re.finditer(r"def\s+(\w+)\s*:\s*(\w+)<(.*?)>\s*;", t, re.S):
+        rec, args = m.group(1), split_top(m.group(3))
+        name = next((a.strip()[1:-1] for a in args if re.fullmatch(r'\s*"[^"]*"\s*', a)), None)
+        if name is None:
+            continue
+        lst = next((a for a in args if a.strip().startswith("[")), "")
+        features.setdefault(rec, (name, re.findall(r"\w+", lst)))
+    # AArch64 architecture levels also carry DEFAULT EXTENSIONS (their last
+    # argument). They are not implied features, but rustc's cfg for
+    # `-C target-cpu` includes them: without them apple-m4 misses `ssbs`
+    # (a v8.5 default) and oryon-1 misses `fp16`/`fhm`, and --verify-rustc
+    # reports exactly those rows.
+    arch_defaults = {m.group(1): split_top(m.group(2))[-1] for m in
+                     re.finditer(r"def\s+(\w+)\s*:\s*Architecture64<(.*?)>\s*;", t, re.S)}
+
+    def defaults(expr):
+        expr = expr.strip()
+        if expr.startswith("!listconcat("):
+            return [r for p in split_top(expr[len("!listconcat("):-1]) for r in defaults(p)]
+        if expr.startswith("!listremove("):
+            base, drop = split_top(expr[len("!listremove("):-1])
+            gone = set(defaults(drop))
+            return [r for r in defaults(base) if r not in gone]
+        if expr.endswith(".DefaultExts"):
+            return defaults(arch_defaults[expr[:-len(".DefaultExts")]])
+        if expr.startswith("["):
+            return re.findall(r"\w+", expr)
+        # Scraping words out of an unknown expression is how `!listremove`
+        # once leaked the architecture def's own name (and its whole implied
+        # chain) into cortex-r82. Refuse instead.
+        sys.exit(f"unsupported TableGen expression in DefaultExts: {expr[:60]}")
+
+    body = re.search(r"def\s+ProcessorFeatures\s*\{(.*?)\n\}", t, re.S).group(1)
+    lists = {m.group(1): m.group(2) for m in
+             re.finditer(r"list<SubtargetFeature>\s+(\w+)\s*=\s*(.*?);", body, re.S)}
+
+    def ev(expr):
+        expr = expr.strip()
+        if expr.startswith("!listconcat("):
+            return [r for p in split_top(expr[len("!listconcat("):-1]) for r in ev(p)]
+        if expr.startswith("!listremove("):
+            base, drop = split_top(expr[len("!listremove("):-1])
+            gone = set(ev(drop))
+            return [r for r in ev(base) if r not in gone]
+        if expr.startswith("!"):
+            sys.exit(f"unsupported TableGen operator in a processor list: {expr[:60]}")
+        if expr.startswith("["):
+            return re.findall(r"\w+", expr)
+        return ev(lists[expr.replace("ProcessorFeatures.", "")])
+
+    def closure(recs):
+        """Transitive implied features. Tune records (`Tune*`, `Tuning*`) are
+        followed but not reported: they are scheduling knobs, yet some imply
+        real ISA features (TuneOryon implies FeatureFullFP16/FP16FML/SPE)."""
+        seen, stack = set(), list(recs)
+        while stack:
+            r = stack.pop()
+            if r in seen or r not in features:
+                continue
+            seen.add(r)
+            stack += features[r][1]
+        return {features[r][0] for r in seen if not r.startswith("Tun")}
+
+    cpus = {}
+    for m in re.finditer(r"def\s*:\s*Proc(?:essor)?Model<\s*\"([^\"]+)\"\s*,\s*\w+\s*,(.*?)>\s*;", t, re.S):
+        args = split_top(m.group(2))
+        recs = ev(args[0])
+        recs += [d for r in recs if r in arch_defaults for d in defaults(arch_defaults[r])]
+        # The tune list (next argument) is applied too, and its records'
+        # implied ISA features are enabled; see closure().
+        if len(args) > 1 and args[1].strip().startswith(("[", "!", "ProcessorFeatures.")):
+            recs += ev(args[1])
+        cpus[m.group(1)] = closure(recs)
+    for m in re.finditer(r'ProcessorAlias<\s*"([^"]+)"\s*,\s*"([^"]+)"', t):
+        if m.group(2) in cpus:
+            cpus[m.group(1)] = cpus[m.group(2)]
+    return cpus
+
+
+def backend(arch, feats):
+    """Mirror of the `simd.rs` cfg ladder (read by hand, not parsed)."""
+    if arch == "x86":
+        if "avx512f" in feats:
+            return "simd_avx512"
+        return "simd_avx2" if "avx2" in feats else "simd_avx2 (no avx2: array polyfill)"
+    return "simd_neon" + (" +dotprod" if "dotprod" in feats else "")
+
+
+def check_primitive_sources(src):
+    for prim, arch, label, insn, gate, _ in PRIMITIVES:
+        for where in (insn, gate):
+            if where and not re.search(where[1], strip_comments(src[where[0]]), re.S):
+                sys.exit(f"LLVM source no longer matches {prim}/{label}: /{where[1]}/ not in {where[0]}")
+
+
+def primitives_for(arch, feats):
+    out = {}
+    for prim in PRIM_ORDER:
+        rows = [r for r in PRIMITIVES if r[0] == prim and r[1] == arch]
+        hit = [r[2] for r in rows if all(n in feats for n in r[5])]
+        out[prim] = hit
+    return out
+
+
+def build(tag, src):
+    check_primitive_sources(src)
+    inv = {"llvm_tag": tag, "arch": {}}
+    for arch in TD:
+        text = "\n".join(src[pathlib.PurePosixPath(f).name] for f in TD[arch]
+                         if "Processors" in f or f.endswith("X86.td") or "Features" in f)
+        cpus = resolve(text)
+        inv["arch"][arch] = {
+            cpu: {"features": sorted(f), "backend": backend(arch, f),
+                  "primitives": primitives_for(arch, f)}
+            for cpu, f in sorted(cpus.items())
+        }
+    return inv
+
+
+def render_md(inv):
+    tag = inv["llvm_tag"]
+    lines = [
+        "# Per-CPU SIMD inventory — generated from LLVM's TableGen source",
+        "",
+        f"> **GENERATED** by `tools/gen_llvm_inventory.py` from llvm-project `{tag}`",
+        "> (the tag matching `rustc -vV` under this repo's pinned toolchain). Do not edit by hand:",
+        "> run `python3 tools/gen_llvm_inventory.py --write`. `--verify-rustc` checks every row",
+        "> against rustc's own `--print cfg -C target-cpu=...`.",
+        "",
+        "`backend` mirrors the `simd.rs` cfg ladder. Primitive columns list the instruction(s)",
+        "that implement the primitive on that CPU; `—` means the CPU has none and a portable",
+        "fallback is required.",
+        "",
+    ]
+    for arch, cpus in inv["arch"].items():
+        shown = SHOWN[arch]
+        lines += [f"## {arch} ({len(cpus)} CPUs)", ""]
+        head = ["cpu", "backend"] + shown + PRIM_ORDER
+        lines.append("| " + " | ".join(head) + " |")
+        lines.append("|" + "---|" * len(head))
+        for cpu, row in cpus.items():
+            feats = set(row["features"])
+            cells = [f"`{cpu}`", row["backend"]] + ["✓" if f in feats else "" for f in shown]
+            cells += ["<br>".join(row["primitives"][p]) or "—" for p in PRIM_ORDER]
+            lines.append("| " + " | ".join(cells) + " |")
+        lines.append("")
+    return "\n".join(lines)
+
+
+def rust_to_llvm_names(llvm_util_rs):
+    """rustc's own Rust-name -> LLVM-name table, parsed from `to_llvm_features`."""
+    body = llvm_util_rs[llvm_util_rs.index("fn to_llvm_features"):]
+    table = {}
+    for arch, marker in (("aarch64", "Arch::AArch64"), ("x86", "Arch::X86")):
+        start = body.index(marker)
+        end = body.index("Arch::", start + len(marker) + 20)
+        section = body[start:end]
+        table[arch] = {r: l for r, l in re.findall(
+            r'"([^"]+)"\s*=>\s*Some\(LLVMFeature::(?:new|with_dependencies)\(\s*"([^"]+)"', section)}
+    return table
+
+
+def verify_rustc(inv, names):
+    """Compare every CPU with rustc's `--print cfg`, over the features stable rustc prints."""
+    bad = 0
+    for arch, cpus in inv["arch"].items():
+        to_rust = {}
+        for r, l in names[arch].items():
+            to_rust.setdefault(l, set()).add(r)
+        cfg = {}
+        for cpu in cpus:
+            try:
+                out = rustc("--print", "cfg", "--target", RUST_TARGET[arch], "-C", f"target-cpu={cpu}")
+            except subprocess.CalledProcessError:
+                print(f"  {arch}/{cpu}: rustc rejects this cpu name (skipped)")
+                continue
+            cfg[cpu] = set(re.findall(r'target_feature="([^"]+)"', out))
+        exposed = set().union(*cfg.values())      # features stable rustc can print at all
+        # Features rustc prints for EVERY cpu are target-level, not per-cpu:
+        # the x86_64 ABI requires sse/sse2, so rustc reports them even for
+        # LLVM's featureless `generic`. They come from the target, not from
+        # any processor list, so they are added to both sides.
+        baseline = set.intersection(*cfg.values())
+        for cpu, rust in cfg.items():
+            mine = ({r for f in cpus[cpu]["features"] for r in to_rust.get(f, {f})} & exposed) | baseline
+            if mine != rust:
+                bad += 1
+                print(f"  MISMATCH {arch}/{cpu}: llvm-only={sorted(mine - rust)} rustc-only={sorted(rust - mine)}")
+        print(f"{arch}: {len(cfg)} CPUs compared against rustc over {len(exposed)} exposed features")
+    return bad
+
+
+def main():
+    ap = argparse.ArgumentParser(description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter)
+    ap.add_argument("--write", action="store_true", help="write the generated outputs")
+    ap.add_argument("--verify-rustc", action="store_true", help="compare every CPU with rustc's table")
+    ap.add_argument("--tag", help="llvm-project tag (default: from rustc -vV)")
+    ap.add_argument("--llvm-src", help="local llvm-project checkout instead of fetching")
+    a = ap.parse_args()
+
+    tag = a.tag or llvm_tag()
+    inv = build(tag, load(tag, a.llvm_src))
+    js = json.dumps(inv, indent=1, sort_keys=True) + "\n"
+    md = render_md(inv)
+
+    if a.verify_rustc:
+        release = re.search(r"release: (\S+)", rustc("-vV")).group(1)
+        cached = CACHE / f"rust-{release}" / "llvm_util.rs"
+        if not cached.exists():
+            cached.parent.mkdir(parents=True, exist_ok=True)
+            url = ("https://raw.githubusercontent.com/rust-lang/rust/"
+                   f"{release}/compiler/rustc_codegen_llvm/src/llvm_util.rs")
+            with urllib.request.urlopen(url, timeout=60) as r:
+                cached.write_bytes(r.read())
+        bad = verify_rustc(inv, rust_to_llvm_names(cached.read_text()))
+        print("rustc agrees on every compared CPU" if not bad else f"{bad} CPU(s) disagree")
+        if bad:
+            sys.exit(1)
+    if a.write:
+        OUT_JSON.parent.mkdir(parents=True, exist_ok=True)
+        OUT_JSON.write_text(js)
+        OUT_MD.write_text(md)
+        n = {k: len(v) for k, v in inv["arch"].items()}
+        print(f"wrote {OUT_JSON.relative_to(ROOT)} and {OUT_MD.relative_to(ROOT)} ({tag}, {n})")
+    elif not a.verify_rustc:
+        stale = [p for p, want in ((OUT_JSON, js), (OUT_MD, md))
+                 if not p.exists() or p.read_text() != want]
+        if stale:
+            sys.exit("stale: " + ", ".join(str(p.relative_to(ROOT)) for p in stale)
+                     + " (run with --write)")
+        print(f"inventory is current ({tag})")
+
+
+if __name__ == "__main__":
+    main()
diff --git a/tools/llvm-inventory/inventory.json b/tools/llvm-inventory/inventory.json
new file mode 100644
index 00000000..7c703690
--- /dev/null
+++ b/tools/llvm-inventory/inventory.json
@@ -0,0 +1,12484 @@
+{
+ "arch": {
+  "aarch64": {
+   "a64fx": {
+    "backend": "simd_neon",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "aggressive-fma",
+     "arith-bcc-fusion",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "disable-unpredicated-ld-st-lower",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "fullfp16",
+     "lor",
+     "lse",
+     "max-interleave-factor-4",
+     "neon",
+     "pan",
+     "pan-rwv",
+     "perfmon",
+     "predictable-select-expensive",
+     "ras",
+     "rdm",
+     "sha2",
+     "store-pair-suppress",
+     "sve",
+     "uaops",
+     "use-postra-scheduler",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "ampere1": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "aggressive-fma",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "amvs",
+     "arith-bcc-fusion",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "cmp-bcc-fusion",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fptoint",
+     "fullfp16",
+     "fuse-address",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-literals",
+     "i8mm",
+     "jsconv",
+     "ldp-aligned-only",
+     "lor",
+     "lse",
+     "lse2",
+     "max-interleave-factor-4",
+     "mpam",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predres",
+     "rand",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "sha2",
+     "sha3",
+     "specrestrict",
+     "ssbs",
+     "store-pair-suppress",
+     "stp-aligned-only",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "use-postra-scheduler",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "ampere1a": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "aggressive-fma",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "amvs",
+     "arith-bcc-fusion",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "cmp-bcc-fusion",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fptoint",
+     "fullfp16",
+     "fuse-address",
+     "fuse-addsub-2reg-const1",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-literals",
+     "i8mm",
+     "jsconv",
+     "ldp-aligned-only",
+     "lor",
+     "lse",
+     "lse2",
+     "max-interleave-factor-4",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predres",
+     "rand",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "sha2",
+     "sha3",
+     "sm4",
+     "specrestrict",
+     "ssbs",
+     "store-pair-suppress",
+     "stp-aligned-only",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "use-postra-scheduler",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "ampere1b": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "aggressive-fma",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "amvs",
+     "arith-bcc-fusion",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "cmp-bcc-fusion",
+     "complxnum",
+     "crc",
+     "cssc",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fptoint",
+     "fullfp16",
+     "fuse-address",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-literals",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "ldp-aligned-only",
+     "lor",
+     "lse",
+     "lse2",
+     "max-interleave-factor-4",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "predres",
+     "rand",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "sha2",
+     "sha3",
+     "sm4",
+     "spe-eef",
+     "specrestrict",
+     "ssbs",
+     "store-pair-suppress",
+     "stp-aligned-only",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "use-postra-scheduler",
+     "vh",
+     "wfxt",
+     "xs"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "ampere1c": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "aggressive-fma",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "amvs",
+     "arith-bcc-fusion",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "cmp-bcc-fusion",
+     "complxnum",
+     "crc",
+     "cssc",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "faminmax",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fp8",
+     "fp8fma",
+     "fptoint",
+     "fullfp16",
+     "fuse-address",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-literals",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "lut",
+     "max-interleave-factor-4",
+     "mec",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "predres",
+     "rand",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "rme",
+     "sb",
+     "sel2",
+     "sha2",
+     "sha3",
+     "sm4",
+     "spe-eef",
+     "specrestrict",
+     "ssbs",
+     "store-pair-suppress",
+     "sve",
+     "sve-aes",
+     "sve-b16b16",
+     "sve-sha3",
+     "sve-sm4",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "use-postra-scheduler",
+     "vh",
+     "wfxt",
+     "xs"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-a10": {
+    "backend": "simd_neon",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "crc",
+     "disable-latency-sched-heuristic",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "fuse-aes",
+     "fuse-crypto-eor",
+     "lor",
+     "neon",
+     "no-zcz-fpr64",
+     "pan",
+     "perfmon",
+     "rdm",
+     "sha2",
+     "store-pair-suppress",
+     "vh",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-a11": {
+    "backend": "simd_neon",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "ccpp",
+     "crc",
+     "disable-latency-sched-heuristic",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "fullfp16",
+     "fuse-aes",
+     "fuse-crypto-eor",
+     "lor",
+     "lse",
+     "neon",
+     "no-zcz-fpr64",
+     "pan",
+     "pan-rwv",
+     "perfmon",
+     "ras",
+     "rdm",
+     "sha2",
+     "store-pair-suppress",
+     "uaops",
+     "vh",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-a12": {
+    "backend": "simd_neon",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "disable-latency-sched-heuristic",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "fullfp16",
+     "fuse-aes",
+     "fuse-crypto-eor",
+     "jsconv",
+     "lor",
+     "lse",
+     "neon",
+     "no-zcz-fpr64",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "ras",
+     "rcpc",
+     "rdm",
+     "sha2",
+     "store-pair-suppress",
+     "uaops",
+     "vh",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-a13": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "am",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "disable-latency-sched-heuristic",
+     "dit",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fullfp16",
+     "fuse-aes",
+     "fuse-crypto-eor",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mpam",
+     "neon",
+     "no-zcz-fpr64",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sel2",
+     "sha2",
+     "sha3",
+     "store-pair-suppress",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "vh",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-a14": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "aggressive-fma",
+     "alternate-sextload-cvt-f32-pattern",
+     "altnzcv",
+     "am",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "disable-latency-sched-heuristic",
+     "dit",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fptoint",
+     "fullfp16",
+     "fuse-address",
+     "fuse-aes",
+     "fuse-arith-logic",
+     "fuse-crypto-eor",
+     "fuse-csel",
+     "fuse-literals",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "max-interleave-factor-4",
+     "mpam",
+     "neon",
+     "no-zcz-fpr64",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "sha2",
+     "sha3",
+     "specrestrict",
+     "ssbs",
+     "store-pair-suppress",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "vh",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-a15": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "altnzcv",
+     "am",
+     "amvs",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "disable-latency-sched-heuristic",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-address",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-arith-logic",
+     "fuse-crypto-eor",
+     "fuse-csel",
+     "fuse-literals",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "max-interleave-factor-4",
+     "mpam",
+     "neon",
+     "no-zcz-fpr64",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "sha2",
+     "sha3",
+     "specrestrict",
+     "ssbs",
+     "store-pair-suppress",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "vh",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-a16": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "altnzcv",
+     "am",
+     "amvs",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "disable-latency-sched-heuristic",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-address",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-arith-logic",
+     "fuse-crypto-eor",
+     "fuse-csel",
+     "fuse-literals",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "max-interleave-factor-4",
+     "mpam",
+     "neon",
+     "no-zcz-fpr64",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "sha2",
+     "sha3",
+     "specrestrict",
+     "ssbs",
+     "store-pair-suppress",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "vh",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-a17": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "altnzcv",
+     "am",
+     "amvs",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "disable-latency-sched-heuristic",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-address",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-arith-logic",
+     "fuse-crypto-eor",
+     "fuse-csel",
+     "fuse-literals",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "max-interleave-factor-4",
+     "mpam",
+     "neon",
+     "no-zcz-fpr64",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "sha2",
+     "sha3",
+     "specrestrict",
+     "ssbs",
+     "store-pair-suppress",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "vh",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-a18": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "altnzcv",
+     "am",
+     "amvs",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "disable-latency-sched-heuristic",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-address",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-arith-logic",
+     "fuse-crypto-eor",
+     "fuse-csel",
+     "fuse-literals",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "max-interleave-factor-4",
+     "mpam",
+     "neon",
+     "no-zcz-fpr64",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "sha2",
+     "sha3",
+     "sme",
+     "sme-f64f64",
+     "sme-i16i64",
+     "sme2",
+     "spe-eef",
+     "specrestrict",
+     "ssbs",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "vh",
+     "wfxt",
+     "xs",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-a19": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "altnzcv",
+     "am",
+     "amvs",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "cssc",
+     "disable-latency-sched-heuristic",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-address",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-arith-logic",
+     "fuse-crypto-eor",
+     "fuse-csel",
+     "fuse-literals",
+     "hbc",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "max-interleave-factor-4",
+     "mpam",
+     "mte",
+     "neon",
+     "no-zcz-fpr64",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "sha2",
+     "sha3",
+     "sme",
+     "sme-b16b16",
+     "sme-f16f16",
+     "sme-f64f64",
+     "sme-i16i64",
+     "sme2",
+     "sme2p1",
+     "spe-eef",
+     "specres2",
+     "specrestrict",
+     "ssbs",
+     "sve-b16b16",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "vh",
+     "wfxt",
+     "xs",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-a7": {
+    "backend": "simd_neon",
+    "features": [
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "disable-latency-sched-heuristic",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "fuse-aes",
+     "fuse-crypto-eor",
+     "neon",
+     "no-zcz-fpr64",
+     "perfmon",
+     "sha2",
+     "store-pair-suppress",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fp-workaround",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-a8": {
+    "backend": "simd_neon",
+    "features": [
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "disable-latency-sched-heuristic",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "fuse-aes",
+     "fuse-crypto-eor",
+     "neon",
+     "no-zcz-fpr64",
+     "perfmon",
+     "sha2",
+     "store-pair-suppress",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fp-workaround",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-a9": {
+    "backend": "simd_neon",
+    "features": [
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "disable-latency-sched-heuristic",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "fuse-aes",
+     "fuse-crypto-eor",
+     "neon",
+     "no-zcz-fpr64",
+     "perfmon",
+     "sha2",
+     "store-pair-suppress",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fp-workaround",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-latest": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "altnzcv",
+     "am",
+     "amvs",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "cssc",
+     "disable-latency-sched-heuristic",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-address",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-arith-logic",
+     "fuse-crypto-eor",
+     "fuse-csel",
+     "fuse-literals",
+     "hbc",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "max-interleave-factor-4",
+     "mpam",
+     "mte",
+     "neon",
+     "no-zcz-fpr64",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "sha2",
+     "sha3",
+     "sme",
+     "sme-b16b16",
+     "sme-f16f16",
+     "sme-f64f64",
+     "sme-i16i64",
+     "sme2",
+     "sme2p1",
+     "spe-eef",
+     "specres2",
+     "specrestrict",
+     "ssbs",
+     "sve-b16b16",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "vh",
+     "wfxt",
+     "xs",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-m1": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "aggressive-fma",
+     "alternate-sextload-cvt-f32-pattern",
+     "altnzcv",
+     "am",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "disable-latency-sched-heuristic",
+     "dit",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fptoint",
+     "fullfp16",
+     "fuse-address",
+     "fuse-aes",
+     "fuse-arith-logic",
+     "fuse-crypto-eor",
+     "fuse-csel",
+     "fuse-literals",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "max-interleave-factor-4",
+     "mpam",
+     "neon",
+     "no-zcz-fpr64",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "sha2",
+     "sha3",
+     "specrestrict",
+     "ssbs",
+     "store-pair-suppress",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "vh",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-m2": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "altnzcv",
+     "am",
+     "amvs",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "disable-latency-sched-heuristic",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-address",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-arith-logic",
+     "fuse-crypto-eor",
+     "fuse-csel",
+     "fuse-literals",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "max-interleave-factor-4",
+     "mpam",
+     "neon",
+     "no-zcz-fpr64",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "sha2",
+     "sha3",
+     "specrestrict",
+     "ssbs",
+     "store-pair-suppress",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "vh",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-m3": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "altnzcv",
+     "am",
+     "amvs",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "disable-latency-sched-heuristic",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-address",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-arith-logic",
+     "fuse-crypto-eor",
+     "fuse-csel",
+     "fuse-literals",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "max-interleave-factor-4",
+     "mpam",
+     "neon",
+     "no-zcz-fpr64",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "sha2",
+     "sha3",
+     "specrestrict",
+     "ssbs",
+     "store-pair-suppress",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "vh",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-m4": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "altnzcv",
+     "am",
+     "amvs",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "disable-latency-sched-heuristic",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-address",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-arith-logic",
+     "fuse-crypto-eor",
+     "fuse-csel",
+     "fuse-literals",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "max-interleave-factor-4",
+     "mpam",
+     "neon",
+     "no-zcz-fpr64",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "sha2",
+     "sha3",
+     "sme",
+     "sme-f64f64",
+     "sme-i16i64",
+     "sme2",
+     "spe-eef",
+     "specrestrict",
+     "ssbs",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "vh",
+     "wfxt",
+     "xs",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-m5": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "altnzcv",
+     "am",
+     "amvs",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "cssc",
+     "disable-latency-sched-heuristic",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-address",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-arith-logic",
+     "fuse-crypto-eor",
+     "fuse-csel",
+     "fuse-literals",
+     "hbc",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "max-interleave-factor-4",
+     "mpam",
+     "mte",
+     "neon",
+     "no-zcz-fpr64",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "sha2",
+     "sha3",
+     "sme",
+     "sme-b16b16",
+     "sme-f16f16",
+     "sme-f64f64",
+     "sme-i16i64",
+     "sme2",
+     "sme2p1",
+     "spe-eef",
+     "specres2",
+     "specrestrict",
+     "ssbs",
+     "sve-b16b16",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "vh",
+     "wfxt",
+     "xs",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-s10": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "altnzcv",
+     "am",
+     "amvs",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "disable-latency-sched-heuristic",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-address",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-arith-logic",
+     "fuse-crypto-eor",
+     "fuse-csel",
+     "fuse-literals",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "max-interleave-factor-4",
+     "mpam",
+     "neon",
+     "no-zcz-fpr64",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "sha2",
+     "sha3",
+     "specrestrict",
+     "ssbs",
+     "store-pair-suppress",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "vh",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-s4": {
+    "backend": "simd_neon",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "disable-latency-sched-heuristic",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "fullfp16",
+     "fuse-aes",
+     "fuse-crypto-eor",
+     "jsconv",
+     "lor",
+     "lse",
+     "neon",
+     "no-zcz-fpr64",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "ras",
+     "rcpc",
+     "rdm",
+     "sha2",
+     "store-pair-suppress",
+     "uaops",
+     "vh",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-s5": {
+    "backend": "simd_neon",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "disable-latency-sched-heuristic",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "fullfp16",
+     "fuse-aes",
+     "fuse-crypto-eor",
+     "jsconv",
+     "lor",
+     "lse",
+     "neon",
+     "no-zcz-fpr64",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "ras",
+     "rcpc",
+     "rdm",
+     "sha2",
+     "store-pair-suppress",
+     "uaops",
+     "vh",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-s6": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "am",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "disable-latency-sched-heuristic",
+     "dit",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fullfp16",
+     "fuse-aes",
+     "fuse-crypto-eor",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mpam",
+     "neon",
+     "no-zcz-fpr64",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sel2",
+     "sha2",
+     "sha3",
+     "store-pair-suppress",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "vh",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-s7": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "am",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "disable-latency-sched-heuristic",
+     "dit",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fullfp16",
+     "fuse-aes",
+     "fuse-crypto-eor",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mpam",
+     "neon",
+     "no-zcz-fpr64",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sel2",
+     "sha2",
+     "sha3",
+     "store-pair-suppress",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "vh",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-s8": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "am",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "disable-latency-sched-heuristic",
+     "dit",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fullfp16",
+     "fuse-aes",
+     "fuse-crypto-eor",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mpam",
+     "neon",
+     "no-zcz-fpr64",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sel2",
+     "sha2",
+     "sha3",
+     "store-pair-suppress",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "vh",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "apple-s9": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "altnzcv",
+     "am",
+     "amvs",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "disable-latency-sched-heuristic",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-address",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-arith-logic",
+     "fuse-crypto-eor",
+     "fuse-csel",
+     "fuse-literals",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "max-interleave-factor-4",
+     "mpam",
+     "neon",
+     "no-zcz-fpr64",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "sha2",
+     "sha3",
+     "specrestrict",
+     "ssbs",
+     "store-pair-suppress",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "vh",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "c1-nano": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "altnzcv",
+     "am",
+     "amvs",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "chk",
+     "clrbhb",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "ete",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "hbc",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mec",
+     "mops",
+     "mpam",
+     "mte",
+     "neon",
+     "nmi",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rcpc3",
+     "rdm",
+     "rme",
+     "sb",
+     "sel2",
+     "sme",
+     "sme2",
+     "spe-eef",
+     "specres2",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-bitperm",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-fixed-over-scalable-if-equal-cost",
+     "use-postra-scheduler",
+     "use-wzr-to-vec-move",
+     "vh",
+     "wfxt",
+     "xs"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "c1-premium": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "amvs",
+     "avoid-ldapur",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "chk",
+     "clrbhb",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "ete",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-csel",
+     "fuse-cset",
+     "hbc",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mec",
+     "mops",
+     "mpam",
+     "mte",
+     "neon",
+     "nmi",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rcpc3",
+     "rdm",
+     "rme",
+     "sb",
+     "sel2",
+     "sme",
+     "sme2",
+     "spe",
+     "spe-eef",
+     "specres2",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-bitperm",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-fixed-over-scalable-if-equal-cost",
+     "use-postra-scheduler",
+     "vh",
+     "wfxt",
+     "xs"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "c1-pro": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "amvs",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "chk",
+     "clrbhb",
+     "cmp-bcc-fusion",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "ete",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-csel",
+     "fuse-cset",
+     "hbc",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mec",
+     "mops",
+     "mpam",
+     "mte",
+     "neon",
+     "nmi",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rcpc3",
+     "rdm",
+     "rme",
+     "sb",
+     "sel2",
+     "sme",
+     "sme2",
+     "spe",
+     "spe-eef",
+     "specres2",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-bitperm",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-postra-scheduler",
+     "vh",
+     "wfxt",
+     "xs"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "c1-ultra": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "amvs",
+     "avoid-ldapur",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "chk",
+     "clrbhb",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "ete",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-csel",
+     "fuse-cset",
+     "hbc",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mec",
+     "mops",
+     "mpam",
+     "mte",
+     "neon",
+     "nmi",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rcpc3",
+     "rdm",
+     "rme",
+     "sb",
+     "sel2",
+     "sme",
+     "sme2",
+     "spe",
+     "spe-eef",
+     "specres2",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-bitperm",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-fixed-over-scalable-if-equal-cost",
+     "use-postra-scheduler",
+     "vh",
+     "wfxt",
+     "xs"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "carmel": {
+    "backend": "simd_neon",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "ccpp",
+     "crc",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "fullfp16",
+     "lor",
+     "lse",
+     "neon",
+     "pan",
+     "pan-rwv",
+     "ras",
+     "rdm",
+     "sha2",
+     "uaops",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cobalt-100": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "disable-maximize-scalable-bandwidth",
+     "dit",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "ete",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-csel",
+     "fuse-cset",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-bitperm",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-postra-scheduler",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a320": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aggressive-interleaving",
+     "altnzcv",
+     "am",
+     "amvs",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "ete",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mec",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "rme",
+     "sb",
+     "sel2",
+     "spe-eef",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-bitperm",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-fixed-over-scalable-if-equal-cost",
+     "use-postra-scheduler",
+     "use-wzr-to-vec-move",
+     "vh",
+     "wfxt",
+     "xs"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a34": {
+    "backend": "simd_neon",
+    "features": [
+     "a",
+     "aes",
+     "crc",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "neon",
+     "perfmon",
+     "sha2"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a35": {
+    "backend": "simd_neon",
+    "features": [
+     "a",
+     "aes",
+     "crc",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "neon",
+     "perfmon",
+     "sha2"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a510": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "altnzcv",
+     "am",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "ete",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-bitperm",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-fixed-over-scalable-if-equal-cost",
+     "use-postra-scheduler",
+     "use-wzr-to-vec-move",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a520": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "altnzcv",
+     "am",
+     "amvs",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "ete",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mec",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "rme",
+     "sb",
+     "sel2",
+     "spe-eef",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-bitperm",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-fixed-over-scalable-if-equal-cost",
+     "use-postra-scheduler",
+     "use-wzr-to-vec-move",
+     "vh",
+     "wfxt",
+     "xs"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a520ae": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "altnzcv",
+     "am",
+     "amvs",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "ete",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mec",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "rme",
+     "sb",
+     "sel2",
+     "spe-eef",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-bitperm",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-postra-scheduler",
+     "use-wzr-to-vec-move",
+     "vh",
+     "wfxt",
+     "xs"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a53": {
+    "backend": "simd_neon",
+    "features": [
+     "a",
+     "aes",
+     "balance-fp-ops",
+     "crc",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "neon",
+     "perfmon",
+     "sha2",
+     "use-postra-scheduler",
+     "use-wzr-to-vec-move"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a55": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "ccpp",
+     "crc",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "fullfp16",
+     "fuse-address",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "lor",
+     "lse",
+     "neon",
+     "pan",
+     "pan-rwv",
+     "perfmon",
+     "ras",
+     "rcpc",
+     "rdm",
+     "sha2",
+     "uaops",
+     "use-postra-scheduler",
+     "use-wzr-to-vec-move",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a57": {
+    "backend": "simd_neon",
+    "features": [
+     "a",
+     "addr-lsl-slow-14",
+     "aes",
+     "balance-fp-ops",
+     "crc",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "fp-armv8",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-literals",
+     "max-interleave-factor-4",
+     "neon",
+     "perfmon",
+     "predictable-select-expensive",
+     "sha2",
+     "use-postra-scheduler"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a65": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "ccpp",
+     "crc",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "fp-armv8",
+     "fullfp16",
+     "fuse-address",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-literals",
+     "lor",
+     "lse",
+     "neon",
+     "pan",
+     "pan-rwv",
+     "perfmon",
+     "predictable-select-expensive",
+     "ras",
+     "rcpc",
+     "rdm",
+     "sha2",
+     "ssbs",
+     "uaops",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a65ae": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "ccpp",
+     "crc",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "fp-armv8",
+     "fullfp16",
+     "fuse-address",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-literals",
+     "lor",
+     "lse",
+     "neon",
+     "pan",
+     "pan-rwv",
+     "perfmon",
+     "predictable-select-expensive",
+     "ras",
+     "rcpc",
+     "rdm",
+     "sha2",
+     "ssbs",
+     "uaops",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a710": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "cmp-bcc-fusion",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "ete",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-csel",
+     "fuse-cset",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-bitperm",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-postra-scheduler",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a715": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "cmp-bcc-fusion",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "ete",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-csel",
+     "fuse-cset",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "spe",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-bitperm",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-postra-scheduler",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a72": {
+    "backend": "simd_neon",
+    "features": [
+     "a",
+     "addr-lsl-slow-14",
+     "aes",
+     "crc",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "fp-armv8",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-literals",
+     "neon",
+     "perfmon",
+     "predictable-select-expensive",
+     "sha2"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a720": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "amvs",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "cmp-bcc-fusion",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "ete",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-csel",
+     "fuse-cset",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mec",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "rme",
+     "sb",
+     "sel2",
+     "spe",
+     "spe-eef",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-bitperm",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-postra-scheduler",
+     "vh",
+     "wfxt",
+     "xs"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a720ae": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "amvs",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "cmp-bcc-fusion",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "ete",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-csel",
+     "fuse-cset",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mec",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "rme",
+     "sb",
+     "sel2",
+     "spe",
+     "spe-eef",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-bitperm",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-postra-scheduler",
+     "vh",
+     "wfxt",
+     "xs"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a725": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "amvs",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "cmp-bcc-fusion",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "ete",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-csel",
+     "fuse-cset",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mec",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "rme",
+     "sb",
+     "sel2",
+     "spe",
+     "spe-eef",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-bitperm",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-postra-scheduler",
+     "vh",
+     "wfxt",
+     "xs"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a73": {
+    "backend": "simd_neon",
+    "features": [
+     "a",
+     "addr-lsl-slow-14",
+     "aes",
+     "crc",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "fp-armv8",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "neon",
+     "perfmon",
+     "predictable-select-expensive",
+     "sha2"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a75": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "addr-lsl-slow-14",
+     "aes",
+     "ccpp",
+     "crc",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "fp-armv8",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "lor",
+     "lse",
+     "neon",
+     "pan",
+     "pan-rwv",
+     "perfmon",
+     "predictable-select-expensive",
+     "ras",
+     "rcpc",
+     "rdm",
+     "sha2",
+     "uaops",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a76": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "addr-lsl-slow-14",
+     "aes",
+     "alu-lsl-fast",
+     "ccpp",
+     "crc",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "fp-armv8",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "lor",
+     "lse",
+     "neon",
+     "pan",
+     "pan-rwv",
+     "perfmon",
+     "predictable-select-expensive",
+     "ras",
+     "rcpc",
+     "rdm",
+     "sha2",
+     "ssbs",
+     "uaops",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a76ae": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "addr-lsl-slow-14",
+     "aes",
+     "alu-lsl-fast",
+     "ccpp",
+     "crc",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "fp-armv8",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "lor",
+     "lse",
+     "neon",
+     "pan",
+     "pan-rwv",
+     "perfmon",
+     "predictable-select-expensive",
+     "ras",
+     "rcpc",
+     "rdm",
+     "sha2",
+     "ssbs",
+     "uaops",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a77": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "addr-lsl-slow-14",
+     "aes",
+     "alu-lsl-fast",
+     "ccpp",
+     "cmp-bcc-fusion",
+     "crc",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "fp-armv8",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "lor",
+     "lse",
+     "neon",
+     "pan",
+     "pan-rwv",
+     "perfmon",
+     "predictable-select-expensive",
+     "ras",
+     "rcpc",
+     "rdm",
+     "sha2",
+     "ssbs",
+     "uaops",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a78": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "addr-lsl-slow-14",
+     "aes",
+     "alu-lsl-fast",
+     "ccpp",
+     "cmp-bcc-fusion",
+     "crc",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "fp-armv8",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-csel",
+     "fuse-cset",
+     "lor",
+     "lse",
+     "neon",
+     "pan",
+     "pan-rwv",
+     "perfmon",
+     "predictable-select-expensive",
+     "ras",
+     "rcpc",
+     "rdm",
+     "sha2",
+     "spe",
+     "ssbs",
+     "uaops",
+     "use-postra-scheduler",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a78ae": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "addr-lsl-slow-14",
+     "aes",
+     "alu-lsl-fast",
+     "ccpp",
+     "cmp-bcc-fusion",
+     "crc",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "fp-armv8",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-csel",
+     "fuse-cset",
+     "lor",
+     "lse",
+     "neon",
+     "pan",
+     "pan-rwv",
+     "perfmon",
+     "predictable-select-expensive",
+     "ras",
+     "rcpc",
+     "rdm",
+     "sha2",
+     "spe",
+     "ssbs",
+     "uaops",
+     "use-postra-scheduler",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-a78c": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "addr-lsl-slow-14",
+     "aes",
+     "alu-lsl-fast",
+     "ccpp",
+     "cmp-bcc-fusion",
+     "crc",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "flagm",
+     "fp-armv8",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-csel",
+     "fuse-cset",
+     "lor",
+     "lse",
+     "neon",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "ras",
+     "rcpc",
+     "rdm",
+     "sha2",
+     "spe",
+     "ssbs",
+     "uaops",
+     "use-postra-scheduler",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-r82": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fullfp16",
+     "jsconv",
+     "lse",
+     "neon",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predres",
+     "r",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "specrestrict",
+     "ssbs",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "use-postra-scheduler"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-r82ae": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fullfp16",
+     "jsconv",
+     "lse",
+     "neon",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predres",
+     "r",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "specrestrict",
+     "ssbs",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "use-postra-scheduler"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-x1": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "addr-lsl-slow-14",
+     "aes",
+     "alu-lsl-fast",
+     "ccpp",
+     "cmp-bcc-fusion",
+     "crc",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "fp-armv8",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "lor",
+     "lse",
+     "neon",
+     "pan",
+     "pan-rwv",
+     "perfmon",
+     "predictable-select-expensive",
+     "ras",
+     "rcpc",
+     "rdm",
+     "sha2",
+     "spe",
+     "ssbs",
+     "uaops",
+     "use-postra-scheduler",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-x1c": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "addr-lsl-slow-14",
+     "aes",
+     "alu-lsl-fast",
+     "ccpp",
+     "cmp-bcc-fusion",
+     "crc",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "flagm",
+     "fp-armv8",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "lor",
+     "lse",
+     "lse2",
+     "neon",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sha2",
+     "spe",
+     "ssbs",
+     "uaops",
+     "use-postra-scheduler",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-x2": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "cmp-bcc-fusion",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "ete",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-bitperm",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-fixed-over-scalable-if-equal-cost",
+     "use-postra-scheduler",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-x3": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "avoid-ldapur",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "ete",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "spe",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-bitperm",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-fixed-over-scalable-if-equal-cost",
+     "use-postra-scheduler",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-x4": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "amvs",
+     "avoid-ldapur",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "ete",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-csel",
+     "fuse-cset",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mec",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "rme",
+     "sb",
+     "sel2",
+     "spe",
+     "spe-eef",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-bitperm",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-fixed-over-scalable-if-equal-cost",
+     "use-postra-scheduler",
+     "vh",
+     "wfxt",
+     "xs"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cortex-x925": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "amvs",
+     "avoid-ldapur",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "ete",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-csel",
+     "fuse-cset",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mec",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "rme",
+     "sb",
+     "sel2",
+     "spe",
+     "spe-eef",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-bitperm",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-fixed-over-scalable-if-equal-cost",
+     "use-postra-scheduler",
+     "vh",
+     "wfxt",
+     "xs"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "cyclone": {
+    "backend": "simd_neon",
+    "features": [
+     "a",
+     "aes",
+     "alternate-sextload-cvt-f32-pattern",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "disable-latency-sched-heuristic",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "fuse-aes",
+     "fuse-crypto-eor",
+     "neon",
+     "no-zcz-fpr64",
+     "perfmon",
+     "sha2",
+     "store-pair-suppress",
+     "zcm-fpr128",
+     "zcm-gpr64",
+     "zcz-fp-workaround",
+     "zcz-fpr128",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "exynos-m3": {
+    "backend": "simd_neon",
+    "features": [
+     "a",
+     "aes",
+     "alu-lsl-fast",
+     "crc",
+     "el2vmsa",
+     "el3",
+     "exynos-cheap-as-move",
+     "force-32bit-jump-tables",
+     "fp-armv8",
+     "fuse-address",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-csel",
+     "fuse-literals",
+     "max-interleave-factor-4",
+     "neon",
+     "perfmon",
+     "predictable-select-expensive",
+     "sha2",
+     "store-pair-suppress",
+     "use-postra-scheduler"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "exynos-m4": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alu-lsl-fast",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "ccpp",
+     "crc",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "exynos-cheap-as-move",
+     "force-32bit-jump-tables",
+     "fp-armv8",
+     "fullfp16",
+     "fuse-address",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-arith-logic",
+     "fuse-csel",
+     "fuse-literals",
+     "lor",
+     "lse",
+     "max-interleave-factor-4",
+     "neon",
+     "pan",
+     "pan-rwv",
+     "perfmon",
+     "ras",
+     "rdm",
+     "sha2",
+     "store-pair-suppress",
+     "uaops",
+     "use-postra-scheduler",
+     "vh",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "exynos-m5": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alu-lsl-fast",
+     "arith-bcc-fusion",
+     "arith-cbz-fusion",
+     "ccpp",
+     "crc",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "exynos-cheap-as-move",
+     "force-32bit-jump-tables",
+     "fp-armv8",
+     "fullfp16",
+     "fuse-address",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-arith-logic",
+     "fuse-csel",
+     "fuse-literals",
+     "lor",
+     "lse",
+     "max-interleave-factor-4",
+     "neon",
+     "pan",
+     "pan-rwv",
+     "perfmon",
+     "ras",
+     "rdm",
+     "sha2",
+     "store-pair-suppress",
+     "uaops",
+     "use-postra-scheduler",
+     "vh",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "falkor": {
+    "backend": "simd_neon",
+    "features": [
+     "a",
+     "aes",
+     "alu-lsl-fast",
+     "crc",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "max-interleave-factor-4",
+     "neon",
+     "perfmon",
+     "predictable-select-expensive",
+     "rdm",
+     "sha2",
+     "slow-strqro-store",
+     "store-pair-suppress",
+     "use-postra-scheduler",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "fujitsu-monaka": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "altnzcv",
+     "am",
+     "amvs",
+     "arith-bcc-fusion",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "clrbhb",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "ete",
+     "faminmax",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fp8",
+     "fp8dot2",
+     "fp8dot4",
+     "fp8fma",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "hbc",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "ls64",
+     "lse",
+     "lse2",
+     "lut",
+     "mec",
+     "mops",
+     "mpam",
+     "neon",
+     "nmi",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "predres",
+     "rand",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "rme",
+     "sb",
+     "sel2",
+     "sha2",
+     "sha3",
+     "sm4",
+     "spe-eef",
+     "specres2",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-aes",
+     "sve-bitperm",
+     "sve-sha3",
+     "sve-sm4",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-postra-scheduler",
+     "vh",
+     "wfxt",
+     "xs"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "gb10": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "amvs",
+     "avoid-ldapur",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "ete",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-csel",
+     "fuse-cset",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mec",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "rme",
+     "sb",
+     "sel2",
+     "sha2",
+     "sha3",
+     "sm4",
+     "spe",
+     "spe-eef",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-aes",
+     "sve-bitperm",
+     "sve-sha3",
+     "sve-sm4",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-fixed-over-scalable-if-equal-cost",
+     "use-postra-scheduler",
+     "vh",
+     "wfxt",
+     "xs"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "generic": {
+    "backend": "simd_neon",
+    "features": [
+     "enable-select-opt",
+     "ete",
+     "fp-armv8",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "neon",
+     "trbe",
+     "use-postra-scheduler"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "grace": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "avoid-ldapur",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "cmp-bcc-fusion",
+     "complxnum",
+     "crc",
+     "disable-latency-sched-heuristic",
+     "dit",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "ete",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-csel",
+     "fuse-cset",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "max-interleave-factor-4",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "sha2",
+     "sha3",
+     "sm4",
+     "spe",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-aes",
+     "sve-bitperm",
+     "sve-sha3",
+     "sve-sm4",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-fixed-over-scalable-if-equal-cost",
+     "use-postra-scheduler",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "kryo": {
+    "backend": "simd_neon",
+    "features": [
+     "a",
+     "aes",
+     "alu-lsl-fast",
+     "crc",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "max-interleave-factor-4",
+     "neon",
+     "perfmon",
+     "predictable-select-expensive",
+     "sha2",
+     "store-pair-suppress",
+     "use-postra-scheduler",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "neoverse-512tvb": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alu-lsl-fast",
+     "am",
+     "bf16",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "max-interleave-factor-4",
+     "mpam",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "rand",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sel2",
+     "sha2",
+     "sha3",
+     "sm4",
+     "spe",
+     "ssbs",
+     "sve",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "use-postra-scheduler",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "neoverse-e1": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "ccpp",
+     "crc",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "lor",
+     "lse",
+     "neon",
+     "pan",
+     "pan-rwv",
+     "perfmon",
+     "ras",
+     "rcpc",
+     "rdm",
+     "sha2",
+     "ssbs",
+     "uaops",
+     "use-postra-scheduler",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "neoverse-n1": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "addr-lsl-slow-14",
+     "aes",
+     "alu-lsl-fast",
+     "ccpp",
+     "crc",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "fp-armv8",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "lor",
+     "lse",
+     "neon",
+     "pan",
+     "pan-rwv",
+     "perfmon",
+     "predictable-select-expensive",
+     "ras",
+     "rcpc",
+     "rdm",
+     "sha2",
+     "spe",
+     "ssbs",
+     "uaops",
+     "use-postra-scheduler",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "neoverse-n2": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "disable-maximize-scalable-bandwidth",
+     "dit",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "ete",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-csel",
+     "fuse-cset",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "predres",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-bitperm",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-postra-scheduler",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "neoverse-n3": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "amvs",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "ete",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-csel",
+     "fuse-cset",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mec",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "predres",
+     "rand",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "rme",
+     "sb",
+     "sel2",
+     "spe",
+     "spe-eef",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-bitperm",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-postra-scheduler",
+     "vh",
+     "wfxt",
+     "xs"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "neoverse-v1": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "addr-lsl-slow-14",
+     "aes",
+     "alu-lsl-fast",
+     "am",
+     "bf16",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "disable-maximize-scalable-bandwidth",
+     "dit",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-csel",
+     "fuse-cset",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "mpam",
+     "neon",
+     "no-sve-fp-ld1r",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "rand",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sel2",
+     "sha2",
+     "sha3",
+     "sm4",
+     "spe",
+     "ssbs",
+     "sve",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "use-postra-scheduler",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "neoverse-v2": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "avoid-ldapur",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "cmp-bcc-fusion",
+     "complxnum",
+     "crc",
+     "disable-latency-sched-heuristic",
+     "dit",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "ete",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-csel",
+     "fuse-cset",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "max-interleave-factor-4",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "predres",
+     "rand",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "spe",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-bitperm",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-fixed-over-scalable-if-equal-cost",
+     "use-postra-scheduler",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "neoverse-v3": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "amvs",
+     "avoid-ldapur",
+     "bf16",
+     "brbe",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "ete",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-csel",
+     "fuse-cset",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "ls64",
+     "lse",
+     "lse2",
+     "max-interleave-factor-4",
+     "mec",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "predres",
+     "rand",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "rme",
+     "sb",
+     "sel2",
+     "spe",
+     "spe-eef",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-bitperm",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-postra-scheduler",
+     "vh",
+     "wfxt",
+     "xs"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "neoverse-v3ae": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "amvs",
+     "avoid-ldapur",
+     "bf16",
+     "brbe",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "ete",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-csel",
+     "fuse-cset",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "ls64",
+     "lse",
+     "lse2",
+     "max-interleave-factor-4",
+     "mec",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "predres",
+     "rand",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "rme",
+     "sb",
+     "sel2",
+     "spe",
+     "spe-eef",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-bitperm",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-postra-scheduler",
+     "vh",
+     "wfxt",
+     "xs"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "olympus": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "altnzcv",
+     "alu-lsl-fast",
+     "am",
+     "amvs",
+     "bf16",
+     "brbe",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "chk",
+     "cmp-bcc-fusion",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "ete",
+     "faminmax",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fp8",
+     "fp8dot2",
+     "fp8dot4",
+     "fp8fma",
+     "fpac",
+     "fptoint",
+     "fullfp16",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "hcx",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "ls64",
+     "lse",
+     "lse2",
+     "lut",
+     "max-interleave-factor-4",
+     "mec",
+     "mpam",
+     "mte",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "predres",
+     "rand",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "rme",
+     "sb",
+     "sel2",
+     "sha2",
+     "sha3",
+     "sm4",
+     "spe",
+     "spe-eef",
+     "specrestrict",
+     "ssbs",
+     "sve",
+     "sve-aes",
+     "sve-bitperm",
+     "sve-sha3",
+     "sve-sm4",
+     "sve2",
+     "tlb-rmi",
+     "tracev8.4",
+     "trbe",
+     "uaops",
+     "use-fixed-over-scalable-if-equal-cost",
+     "use-postra-scheduler",
+     "vh",
+     "wfxt",
+     "xs"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "oryon-1": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "altnzcv",
+     "am",
+     "amvs",
+     "bf16",
+     "bti",
+     "ccdp",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "ecv",
+     "el2vmsa",
+     "el3",
+     "enable-select-opt",
+     "fgt",
+     "flagm",
+     "fp-armv8",
+     "fp16fml",
+     "fptoint",
+     "fullfp16",
+     "fuse-address",
+     "fuse-adrp-add",
+     "fuse-aes",
+     "fuse-crypto-eor",
+     "i8mm",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "max-interleave-factor-4",
+     "mpam",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predres",
+     "rand",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sb",
+     "sel2",
+     "sha2",
+     "sha3",
+     "sm4",
+     "spe",
+     "specrestrict",
+     "ssbs",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "use-postra-scheduler",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "saphira": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "alu-lsl-fast",
+     "am",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "dit",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "flagm",
+     "fp-armv8",
+     "jsconv",
+     "lor",
+     "lse",
+     "lse2",
+     "max-interleave-factor-4",
+     "mpam",
+     "neon",
+     "nv",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "ras",
+     "rcpc",
+     "rcpc-immo",
+     "rdm",
+     "sel2",
+     "sha2",
+     "spe",
+     "store-pair-suppress",
+     "tlb-rmi",
+     "tracev8.4",
+     "uaops",
+     "use-postra-scheduler",
+     "vh",
+     "zcz-gpr32",
+     "zcz-gpr64"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "thunderx": {
+    "backend": "simd_neon",
+    "features": [
+     "a",
+     "aes",
+     "crc",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "neon",
+     "perfmon",
+     "predictable-select-expensive",
+     "sha2",
+     "store-pair-suppress",
+     "use-postra-scheduler"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "thunderx2t99": {
+    "backend": "simd_neon",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "aggressive-fma",
+     "arith-bcc-fusion",
+     "crc",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "lor",
+     "lse",
+     "max-interleave-factor-4",
+     "neon",
+     "pan",
+     "predictable-select-expensive",
+     "rdm",
+     "sha2",
+     "store-pair-suppress",
+     "use-postra-scheduler",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "thunderx3t110": {
+    "backend": "simd_neon",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "aggressive-fma",
+     "arith-bcc-fusion",
+     "balance-fp-ops",
+     "ccidx",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "jsconv",
+     "lor",
+     "lse",
+     "max-interleave-factor-4",
+     "neon",
+     "pan",
+     "pan-rwv",
+     "pauth",
+     "perfmon",
+     "predictable-select-expensive",
+     "ras",
+     "rcpc",
+     "rdm",
+     "sha2",
+     "store-pair-suppress",
+     "strict-align",
+     "uaops",
+     "use-postra-scheduler",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "thunderxt81": {
+    "backend": "simd_neon",
+    "features": [
+     "a",
+     "aes",
+     "crc",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "neon",
+     "perfmon",
+     "predictable-select-expensive",
+     "sha2",
+     "store-pair-suppress",
+     "use-postra-scheduler"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "thunderxt83": {
+    "backend": "simd_neon",
+    "features": [
+     "a",
+     "aes",
+     "crc",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "neon",
+     "perfmon",
+     "predictable-select-expensive",
+     "sha2",
+     "store-pair-suppress",
+     "use-postra-scheduler"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "thunderxt88": {
+    "backend": "simd_neon",
+    "features": [
+     "a",
+     "aes",
+     "crc",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "neon",
+     "perfmon",
+     "predictable-select-expensive",
+     "sha2",
+     "store-pair-suppress",
+     "use-postra-scheduler"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   },
+   "tsv110": {
+    "backend": "simd_neon +dotprod",
+    "features": [
+     "CONTEXTIDREL2",
+     "a",
+     "aes",
+     "ccpp",
+     "complxnum",
+     "crc",
+     "dotprod",
+     "el2vmsa",
+     "el3",
+     "fp-armv8",
+     "fp16fml",
+     "fullfp16",
+     "fuse-aes",
+     "jsconv",
+     "lor",
+     "lse",
+     "neon",
+     "pan",
+     "pan-rwv",
+     "perfmon",
+     "ras",
+     "rdm",
+     "sha2",
+     "spe",
+     "store-pair-suppress",
+     "uaops",
+     "use-postra-scheduler",
+     "vh"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESE + AESMC"
+     ],
+     "clmul": [
+      "PMULL (64x64->128)"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "UMULL / UMULL2"
+     ],
+     "permute_var_u32": [
+      "TBL"
+     ],
+     "shift_var_u32": [
+      "USHL"
+     ]
+    }
+   }
+  },
+  "x86": {
+   "alderlake": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avxvnni",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "gfni",
+     "hreset",
+     "invpcid",
+     "kl",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "movdir64b",
+     "movdiri",
+     "nopl",
+     "pclmul",
+     "pconfig",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "ptwrite",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "serialize",
+     "sha",
+     "shstk",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "vaes",
+     "vpclmulqdq",
+     "waitpkg",
+     "widekl",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "arrowlake": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avxifma",
+     "avxneconvert",
+     "avxvnni",
+     "avxvnniint8",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "cmpccxadd",
+     "crc32",
+     "cx16",
+     "cx8",
+     "enqcmd",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "gfni",
+     "hreset",
+     "invpcid",
+     "kl",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "movdir64b",
+     "movdiri",
+     "nopl",
+     "pclmul",
+     "pconfig",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "ptwrite",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "serialize",
+     "sha",
+     "shstk",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "uintr",
+     "vaes",
+     "vpclmulqdq",
+     "waitpkg",
+     "widekl",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (VEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "arrowlake-s": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avxifma",
+     "avxneconvert",
+     "avxvnni",
+     "avxvnniint16",
+     "avxvnniint8",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "cmpccxadd",
+     "crc32",
+     "cx16",
+     "cx8",
+     "enqcmd",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "gfni",
+     "hreset",
+     "invpcid",
+     "kl",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "movdir64b",
+     "movdiri",
+     "nopl",
+     "pclmul",
+     "pconfig",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "ptwrite",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "serialize",
+     "sha",
+     "sha512",
+     "shstk",
+     "sm3",
+     "sm4",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "uintr",
+     "vaes",
+     "vpclmulqdq",
+     "waitpkg",
+     "widekl",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (VEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "arrowlake_s": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avxifma",
+     "avxneconvert",
+     "avxvnni",
+     "avxvnniint16",
+     "avxvnniint8",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "cmpccxadd",
+     "crc32",
+     "cx16",
+     "cx8",
+     "enqcmd",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "gfni",
+     "hreset",
+     "invpcid",
+     "kl",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "movdir64b",
+     "movdiri",
+     "nopl",
+     "pclmul",
+     "pconfig",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "ptwrite",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "serialize",
+     "sha",
+     "sha512",
+     "shstk",
+     "sm3",
+     "sm4",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "uintr",
+     "vaes",
+     "vpclmulqdq",
+     "waitpkg",
+     "widekl",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (VEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "atom": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "",
+     "64bit",
+     "cmov",
+     "cx16",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "movbe",
+     "nopl",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "ssse3",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "atom_sse4_2": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "prfchw",
+     "rdrnd",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "atom_sse4_2_movbe": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "aes",
+     "clflushopt",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "fsgsbase",
+     "fxsr",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "prfchw",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sha",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "bdver1": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "aes",
+     "avx",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "fma4",
+     "fxsr",
+     "lwp",
+     "lzcnt",
+     "mmx",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "prfchw",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "sse4a",
+     "ssse3",
+     "x87",
+     "xop",
+     "xsave"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "bdver2": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "aes",
+     "avx",
+     "bmi",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "f16c",
+     "fma",
+     "fma4",
+     "fxsr",
+     "lwp",
+     "lzcnt",
+     "mmx",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "prfchw",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "sse4a",
+     "ssse3",
+     "tbm",
+     "x87",
+     "xop",
+     "xsave"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "bdver3": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "aes",
+     "avx",
+     "bmi",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "f16c",
+     "fma",
+     "fma4",
+     "fsgsbase",
+     "fxsr",
+     "lwp",
+     "lzcnt",
+     "mmx",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "prfchw",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "sse4a",
+     "ssse3",
+     "tbm",
+     "x87",
+     "xop",
+     "xsave",
+     "xsaveopt"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "bdver4": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "aes",
+     "avx",
+     "avx2",
+     "bmi",
+     "bmi2",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "f16c",
+     "fma",
+     "fma4",
+     "fsgsbase",
+     "fxsr",
+     "lwp",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "mwaitx",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "prfchw",
+     "rdrnd",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "sse4a",
+     "ssse3",
+     "tbm",
+     "x87",
+     "xop",
+     "xsave",
+     "xsaveopt"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "bonnell": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "",
+     "64bit",
+     "cmov",
+     "cx16",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "movbe",
+     "nopl",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "ssse3",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "broadwell": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "adx",
+     "avx",
+     "avx2",
+     "bmi",
+     "bmi2",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "prfchw",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsaveopt"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "btver2": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "aes",
+     "avx",
+     "bmi",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "f16c",
+     "fxsr",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "prfchw",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "sse4a",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsaveopt"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "cannonlake": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avx512bw",
+     "avx512cd",
+     "avx512dq",
+     "avx512f",
+     "avx512ifma",
+     "avx512vbmi",
+     "avx512vl",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sha",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (EVEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "cascadelake": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avx512bw",
+     "avx512cd",
+     "avx512dq",
+     "avx512f",
+     "avx512vl",
+     "avx512vnni",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "clearwaterforest": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avxifma",
+     "avxneconvert",
+     "avxvnni",
+     "avxvnniint16",
+     "avxvnniint8",
+     "bmi",
+     "bmi2",
+     "cldemote",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "cmpccxadd",
+     "crc32",
+     "cx16",
+     "cx8",
+     "enqcmd",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "gfni",
+     "hreset",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "movdir64b",
+     "movdiri",
+     "nopl",
+     "pclmul",
+     "pconfig",
+     "pku",
+     "popcnt",
+     "prefetchi",
+     "prfchw",
+     "ptwrite",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "serialize",
+     "sha",
+     "sha512",
+     "shstk",
+     "sm3",
+     "sm4",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "uintr",
+     "usermsr",
+     "vaes",
+     "vpclmulqdq",
+     "waitpkg",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (VEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "cooperlake": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avx512bf16",
+     "avx512bw",
+     "avx512cd",
+     "avx512dq",
+     "avx512f",
+     "avx512vl",
+     "avx512vnni",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "core-avx-i": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "avx",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "f16c",
+     "fsgsbase",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "rdrnd",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsaveopt"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "core-avx2": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "avx",
+     "avx2",
+     "bmi",
+     "bmi2",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "rdrnd",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsaveopt"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "core2": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "cmov",
+     "cx16",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "ssse3",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "core_2_duo_sse4_1": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "cmov",
+     "cx16",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "ssse3",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "core_2_duo_ssse3": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "cmov",
+     "cx16",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "ssse3",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "core_2nd_gen_avx": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "avx",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsaveopt"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "core_3rd_gen_avx": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "avx",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "f16c",
+     "fsgsbase",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "rdrnd",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsaveopt"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "core_4th_gen_avx": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "avx",
+     "avx2",
+     "bmi",
+     "bmi2",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "rdrnd",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsaveopt"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "core_4th_gen_avx_tsx": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "avx",
+     "avx2",
+     "bmi",
+     "bmi2",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "rdrnd",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsaveopt"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "core_5th_gen_avx": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "adx",
+     "avx",
+     "avx2",
+     "bmi",
+     "bmi2",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "prfchw",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsaveopt"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "core_5th_gen_avx_tsx": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "adx",
+     "avx",
+     "avx2",
+     "bmi",
+     "bmi2",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "prfchw",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsaveopt"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "core_aes_pclmulqdq": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "core_i7_sse4_2": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "popcnt",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "corei7": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "popcnt",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "corei7-avx": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "avx",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsaveopt"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "diamondrapids": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "amx-avx512",
+     "amx-bf16",
+     "amx-complex",
+     "amx-fp16",
+     "amx-fp8",
+     "amx-int8",
+     "amx-movrs",
+     "amx-tf32",
+     "amx-tile",
+     "avx",
+     "avx10.1",
+     "avx10.2",
+     "avx2",
+     "avx512bf16",
+     "avx512bitalg",
+     "avx512bw",
+     "avx512cd",
+     "avx512dq",
+     "avx512f",
+     "avx512fp16",
+     "avx512ifma",
+     "avx512vbmi",
+     "avx512vbmi2",
+     "avx512vl",
+     "avx512vnni",
+     "avx512vpopcntdq",
+     "avxifma",
+     "avxneconvert",
+     "avxvnni",
+     "avxvnniint16",
+     "avxvnniint8",
+     "bmi",
+     "bmi2",
+     "ccmp",
+     "cldemote",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "cmpccxadd",
+     "crc32",
+     "cx16",
+     "cx8",
+     "egpr",
+     "enqcmd",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fsrm",
+     "fxsr",
+     "gfni",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "movdir64b",
+     "movdiri",
+     "movrs",
+     "ndd",
+     "nf",
+     "nopl",
+     "pclmul",
+     "pconfig",
+     "pku",
+     "popcnt",
+     "ppx",
+     "prefetchi",
+     "prfchw",
+     "ptwrite",
+     "push2pop2",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "serialize",
+     "sha",
+     "sha512",
+     "shstk",
+     "sm3",
+     "sm4",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "tsxldtrk",
+     "uintr",
+     "vaes",
+     "vpclmulqdq",
+     "waitpkg",
+     "wbnoinvd",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves",
+     "zu"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC",
+      "VAESENC (zmm)"
+     ],
+     "clmul": [
+      "PCLMULQDQ",
+      "VPCLMULQDQ (zmm)"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (EVEX)",
+      "VPMADD52LUQ/HUQ (VEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "emeraldrapids": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "amx-bf16",
+     "amx-int8",
+     "amx-tile",
+     "avx",
+     "avx2",
+     "avx512bf16",
+     "avx512bitalg",
+     "avx512bw",
+     "avx512cd",
+     "avx512dq",
+     "avx512f",
+     "avx512fp16",
+     "avx512ifma",
+     "avx512vbmi",
+     "avx512vbmi2",
+     "avx512vl",
+     "avx512vnni",
+     "avx512vpopcntdq",
+     "avxvnni",
+     "bmi",
+     "bmi2",
+     "cldemote",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "enqcmd",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fsrm",
+     "fxsr",
+     "gfni",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "movdir64b",
+     "movdiri",
+     "nopl",
+     "pclmul",
+     "pconfig",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "ptwrite",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "serialize",
+     "sha",
+     "shstk",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "tsxldtrk",
+     "uintr",
+     "vaes",
+     "vpclmulqdq",
+     "waitpkg",
+     "wbnoinvd",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC",
+      "VAESENC (zmm)"
+     ],
+     "clmul": [
+      "PCLMULQDQ",
+      "VPCLMULQDQ (zmm)"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (EVEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "generic": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "cx8",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "goldmont": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "aes",
+     "clflushopt",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "fsgsbase",
+     "fxsr",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "prfchw",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sha",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "goldmont-plus": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "aes",
+     "clflushopt",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "fsgsbase",
+     "fxsr",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "prfchw",
+     "ptwrite",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sha",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "goldmont_plus": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "aes",
+     "clflushopt",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "fsgsbase",
+     "fxsr",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "prfchw",
+     "ptwrite",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sha",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "gracemont": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avxvnni",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "gfni",
+     "hreset",
+     "invpcid",
+     "kl",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "movdir64b",
+     "movdiri",
+     "nopl",
+     "pclmul",
+     "pconfig",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "ptwrite",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "serialize",
+     "sha",
+     "shstk",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "vaes",
+     "vpclmulqdq",
+     "waitpkg",
+     "widekl",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "grandridge": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avxifma",
+     "avxneconvert",
+     "avxvnni",
+     "avxvnniint8",
+     "bmi",
+     "bmi2",
+     "cldemote",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "cmpccxadd",
+     "crc32",
+     "cx16",
+     "cx8",
+     "enqcmd",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "gfni",
+     "hreset",
+     "invpcid",
+     "kl",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "movdir64b",
+     "movdiri",
+     "nopl",
+     "pclmul",
+     "pconfig",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "ptwrite",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "serialize",
+     "sha",
+     "shstk",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "uintr",
+     "vaes",
+     "vpclmulqdq",
+     "waitpkg",
+     "widekl",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (VEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "graniterapids": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "amx-bf16",
+     "amx-fp16",
+     "amx-int8",
+     "amx-tile",
+     "avx",
+     "avx2",
+     "avx512bf16",
+     "avx512bitalg",
+     "avx512bw",
+     "avx512cd",
+     "avx512dq",
+     "avx512f",
+     "avx512fp16",
+     "avx512ifma",
+     "avx512vbmi",
+     "avx512vbmi2",
+     "avx512vl",
+     "avx512vnni",
+     "avx512vpopcntdq",
+     "avxvnni",
+     "bmi",
+     "bmi2",
+     "cldemote",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "enqcmd",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fsrm",
+     "fxsr",
+     "gfni",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "movdir64b",
+     "movdiri",
+     "nopl",
+     "pclmul",
+     "pconfig",
+     "pku",
+     "popcnt",
+     "prefetchi",
+     "prfchw",
+     "ptwrite",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "serialize",
+     "sha",
+     "shstk",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "tsxldtrk",
+     "uintr",
+     "vaes",
+     "vpclmulqdq",
+     "waitpkg",
+     "wbnoinvd",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC",
+      "VAESENC (zmm)"
+     ],
+     "clmul": [
+      "PCLMULQDQ",
+      "VPCLMULQDQ (zmm)"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (EVEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "graniterapids-d": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "amx-bf16",
+     "amx-complex",
+     "amx-fp16",
+     "amx-int8",
+     "amx-tile",
+     "avx",
+     "avx2",
+     "avx512bf16",
+     "avx512bitalg",
+     "avx512bw",
+     "avx512cd",
+     "avx512dq",
+     "avx512f",
+     "avx512fp16",
+     "avx512ifma",
+     "avx512vbmi",
+     "avx512vbmi2",
+     "avx512vl",
+     "avx512vnni",
+     "avx512vpopcntdq",
+     "avxvnni",
+     "bmi",
+     "bmi2",
+     "cldemote",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "enqcmd",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fsrm",
+     "fxsr",
+     "gfni",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "movdir64b",
+     "movdiri",
+     "nopl",
+     "pclmul",
+     "pconfig",
+     "pku",
+     "popcnt",
+     "prefetchi",
+     "prfchw",
+     "ptwrite",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "serialize",
+     "sha",
+     "shstk",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "tsxldtrk",
+     "uintr",
+     "vaes",
+     "vpclmulqdq",
+     "waitpkg",
+     "wbnoinvd",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC",
+      "VAESENC (zmm)"
+     ],
+     "clmul": [
+      "PCLMULQDQ",
+      "VPCLMULQDQ (zmm)"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (EVEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "graniterapids_d": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "amx-bf16",
+     "amx-complex",
+     "amx-fp16",
+     "amx-int8",
+     "amx-tile",
+     "avx",
+     "avx2",
+     "avx512bf16",
+     "avx512bitalg",
+     "avx512bw",
+     "avx512cd",
+     "avx512dq",
+     "avx512f",
+     "avx512fp16",
+     "avx512ifma",
+     "avx512vbmi",
+     "avx512vbmi2",
+     "avx512vl",
+     "avx512vnni",
+     "avx512vpopcntdq",
+     "avxvnni",
+     "bmi",
+     "bmi2",
+     "cldemote",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "enqcmd",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fsrm",
+     "fxsr",
+     "gfni",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "movdir64b",
+     "movdiri",
+     "nopl",
+     "pclmul",
+     "pconfig",
+     "pku",
+     "popcnt",
+     "prefetchi",
+     "prfchw",
+     "ptwrite",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "serialize",
+     "sha",
+     "shstk",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "tsxldtrk",
+     "uintr",
+     "vaes",
+     "vpclmulqdq",
+     "waitpkg",
+     "wbnoinvd",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC",
+      "VAESENC (zmm)"
+     ],
+     "clmul": [
+      "PCLMULQDQ",
+      "VPCLMULQDQ (zmm)"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (EVEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "haswell": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "avx",
+     "avx2",
+     "bmi",
+     "bmi2",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "rdrnd",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsaveopt"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "icelake-client": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avx512bitalg",
+     "avx512bw",
+     "avx512cd",
+     "avx512dq",
+     "avx512f",
+     "avx512ifma",
+     "avx512vbmi",
+     "avx512vbmi2",
+     "avx512vl",
+     "avx512vnni",
+     "avx512vpopcntdq",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fsrm",
+     "fxsr",
+     "gfni",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sha",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "vaes",
+     "vpclmulqdq",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC",
+      "VAESENC (zmm)"
+     ],
+     "clmul": [
+      "PCLMULQDQ",
+      "VPCLMULQDQ (zmm)"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (EVEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "icelake-server": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avx512bitalg",
+     "avx512bw",
+     "avx512cd",
+     "avx512dq",
+     "avx512f",
+     "avx512ifma",
+     "avx512vbmi",
+     "avx512vbmi2",
+     "avx512vl",
+     "avx512vnni",
+     "avx512vpopcntdq",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fsrm",
+     "fxsr",
+     "gfni",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "pconfig",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sha",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "vaes",
+     "vpclmulqdq",
+     "wbnoinvd",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC",
+      "VAESENC (zmm)"
+     ],
+     "clmul": [
+      "PCLMULQDQ",
+      "VPCLMULQDQ (zmm)"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (EVEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "icelake_client": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avx512bitalg",
+     "avx512bw",
+     "avx512cd",
+     "avx512dq",
+     "avx512f",
+     "avx512ifma",
+     "avx512vbmi",
+     "avx512vbmi2",
+     "avx512vl",
+     "avx512vnni",
+     "avx512vpopcntdq",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fsrm",
+     "fxsr",
+     "gfni",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sha",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "vaes",
+     "vpclmulqdq",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC",
+      "VAESENC (zmm)"
+     ],
+     "clmul": [
+      "PCLMULQDQ",
+      "VPCLMULQDQ (zmm)"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (EVEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "icelake_server": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avx512bitalg",
+     "avx512bw",
+     "avx512cd",
+     "avx512dq",
+     "avx512f",
+     "avx512ifma",
+     "avx512vbmi",
+     "avx512vbmi2",
+     "avx512vl",
+     "avx512vnni",
+     "avx512vpopcntdq",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fsrm",
+     "fxsr",
+     "gfni",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "pconfig",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sha",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "vaes",
+     "vpclmulqdq",
+     "wbnoinvd",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC",
+      "VAESENC (zmm)"
+     ],
+     "clmul": [
+      "PCLMULQDQ",
+      "VPCLMULQDQ (zmm)"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (EVEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "ivybridge": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "avx",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "f16c",
+     "fsgsbase",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "rdrnd",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsaveopt"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "knl": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avx512cd",
+     "avx512f",
+     "bmi",
+     "bmi2",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "prfchw",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsaveopt"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "knm": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avx512cd",
+     "avx512f",
+     "avx512vpopcntdq",
+     "bmi",
+     "bmi2",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "prfchw",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsaveopt"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "lunarlake": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avxifma",
+     "avxneconvert",
+     "avxvnni",
+     "avxvnniint16",
+     "avxvnniint8",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "cmpccxadd",
+     "crc32",
+     "cx16",
+     "cx8",
+     "enqcmd",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "gfni",
+     "hreset",
+     "invpcid",
+     "kl",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "movdir64b",
+     "movdiri",
+     "nopl",
+     "pclmul",
+     "pconfig",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "ptwrite",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "serialize",
+     "sha",
+     "sha512",
+     "shstk",
+     "sm3",
+     "sm4",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "uintr",
+     "vaes",
+     "vpclmulqdq",
+     "waitpkg",
+     "widekl",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (VEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "meteorlake": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avxvnni",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "gfni",
+     "hreset",
+     "invpcid",
+     "kl",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "movdir64b",
+     "movdiri",
+     "nopl",
+     "pclmul",
+     "pconfig",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "ptwrite",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "serialize",
+     "sha",
+     "shstk",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "vaes",
+     "vpclmulqdq",
+     "waitpkg",
+     "widekl",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "mic_avx512": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avx512cd",
+     "avx512f",
+     "bmi",
+     "bmi2",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "prfchw",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsaveopt"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "nehalem": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "popcnt",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "nocona": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "cmov",
+     "cx16",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "sse",
+     "sse2",
+     "sse3",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "novalake": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx10.1",
+     "avx10.2",
+     "avx2",
+     "avx512bf16",
+     "avx512bitalg",
+     "avx512bw",
+     "avx512cd",
+     "avx512dq",
+     "avx512f",
+     "avx512fp16",
+     "avx512ifma",
+     "avx512vbmi",
+     "avx512vbmi2",
+     "avx512vl",
+     "avx512vnni",
+     "avx512vpopcntdq",
+     "avxifma",
+     "avxneconvert",
+     "avxvnni",
+     "avxvnniint16",
+     "avxvnniint8",
+     "bmi",
+     "bmi2",
+     "ccmp",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "cmpccxadd",
+     "crc32",
+     "cx16",
+     "cx8",
+     "egpr",
+     "enqcmd",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "gfni",
+     "hreset",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "movdir64b",
+     "movdiri",
+     "movrs",
+     "ndd",
+     "nf",
+     "nopl",
+     "pclmul",
+     "pconfig",
+     "pku",
+     "popcnt",
+     "ppx",
+     "prefetchi",
+     "prfchw",
+     "ptwrite",
+     "push2pop2",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "serialize",
+     "sha",
+     "sha512",
+     "shstk",
+     "sm3",
+     "sm4",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "uintr",
+     "vaes",
+     "vpclmulqdq",
+     "waitpkg",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves",
+     "zu"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC",
+      "VAESENC (zmm)"
+     ],
+     "clmul": [
+      "PCLMULQDQ",
+      "VPCLMULQDQ (zmm)"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (EVEX)",
+      "VPMADD52LUQ/HUQ (VEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "pantherlake": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avxifma",
+     "avxneconvert",
+     "avxvnni",
+     "avxvnniint16",
+     "avxvnniint8",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "cmpccxadd",
+     "crc32",
+     "cx16",
+     "cx8",
+     "enqcmd",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "gfni",
+     "hreset",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "movdir64b",
+     "movdiri",
+     "nopl",
+     "pclmul",
+     "pconfig",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "ptwrite",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "serialize",
+     "sha",
+     "sha512",
+     "shstk",
+     "sm3",
+     "sm4",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "uintr",
+     "vaes",
+     "vpclmulqdq",
+     "waitpkg",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (VEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "penryn": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "cmov",
+     "cx16",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "ssse3",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "pentium-m": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "cmov",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "sse",
+     "sse2",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "pentium4": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "cmov",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "sse",
+     "sse2",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "pentium4m": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "cmov",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "sse",
+     "sse2",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "pentium_4": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "cmov",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "sse",
+     "sse2",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "pentium_4_sse3": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "cmov",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "sse",
+     "sse2",
+     "sse3",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "pentium_m": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "cmov",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "sse",
+     "sse2",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "prescott": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "cmov",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "sse",
+     "sse2",
+     "sse3",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "raptorlake": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avxvnni",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "gfni",
+     "hreset",
+     "invpcid",
+     "kl",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "movdir64b",
+     "movdiri",
+     "nopl",
+     "pclmul",
+     "pconfig",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "ptwrite",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "serialize",
+     "sha",
+     "shstk",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "vaes",
+     "vpclmulqdq",
+     "waitpkg",
+     "widekl",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "rocketlake": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avx512bitalg",
+     "avx512bw",
+     "avx512cd",
+     "avx512dq",
+     "avx512f",
+     "avx512ifma",
+     "avx512vbmi",
+     "avx512vbmi2",
+     "avx512vl",
+     "avx512vnni",
+     "avx512vpopcntdq",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fsrm",
+     "fxsr",
+     "gfni",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sha",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "vaes",
+     "vpclmulqdq",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC",
+      "VAESENC (zmm)"
+     ],
+     "clmul": [
+      "PCLMULQDQ",
+      "VPCLMULQDQ (zmm)"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (EVEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "sandybridge": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "avx",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsaveopt"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "sapphirerapids": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "amx-bf16",
+     "amx-int8",
+     "amx-tile",
+     "avx",
+     "avx2",
+     "avx512bf16",
+     "avx512bitalg",
+     "avx512bw",
+     "avx512cd",
+     "avx512dq",
+     "avx512f",
+     "avx512fp16",
+     "avx512ifma",
+     "avx512vbmi",
+     "avx512vbmi2",
+     "avx512vl",
+     "avx512vnni",
+     "avx512vpopcntdq",
+     "avxvnni",
+     "bmi",
+     "bmi2",
+     "cldemote",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "enqcmd",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fsrm",
+     "fxsr",
+     "gfni",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "movdir64b",
+     "movdiri",
+     "nopl",
+     "pclmul",
+     "pconfig",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "ptwrite",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "serialize",
+     "sha",
+     "shstk",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "tsxldtrk",
+     "uintr",
+     "vaes",
+     "vpclmulqdq",
+     "waitpkg",
+     "wbnoinvd",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC",
+      "VAESENC (zmm)"
+     ],
+     "clmul": [
+      "PCLMULQDQ",
+      "VPCLMULQDQ (zmm)"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (EVEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "sierraforest": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avxifma",
+     "avxneconvert",
+     "avxvnni",
+     "avxvnniint8",
+     "bmi",
+     "bmi2",
+     "cldemote",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "cmpccxadd",
+     "crc32",
+     "cx16",
+     "cx8",
+     "enqcmd",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "gfni",
+     "hreset",
+     "invpcid",
+     "kl",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "movdir64b",
+     "movdiri",
+     "nopl",
+     "pclmul",
+     "pconfig",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "ptwrite",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "serialize",
+     "sha",
+     "shstk",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "uintr",
+     "vaes",
+     "vpclmulqdq",
+     "waitpkg",
+     "widekl",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (VEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "silvermont": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "prfchw",
+     "rdrnd",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "skx": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avx512bw",
+     "avx512cd",
+     "avx512dq",
+     "avx512f",
+     "avx512vl",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "skylake": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "prfchw",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "skylake-avx512": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avx512bw",
+     "avx512cd",
+     "avx512dq",
+     "avx512f",
+     "avx512vl",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "skylake_avx512": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avx512bw",
+     "avx512cd",
+     "avx512dq",
+     "avx512f",
+     "avx512vl",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "slm": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "prfchw",
+     "rdrnd",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "tigerlake": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avx512bitalg",
+     "avx512bw",
+     "avx512cd",
+     "avx512dq",
+     "avx512f",
+     "avx512ifma",
+     "avx512vbmi",
+     "avx512vbmi2",
+     "avx512vl",
+     "avx512vnni",
+     "avx512vp2intersect",
+     "avx512vpopcntdq",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "ermsb",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fsrm",
+     "fxsr",
+     "gfni",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "movdir64b",
+     "movdiri",
+     "nopl",
+     "pclmul",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sha",
+     "shstk",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "vaes",
+     "vpclmulqdq",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC",
+      "VAESENC (zmm)"
+     ],
+     "clmul": [
+      "PCLMULQDQ",
+      "VPCLMULQDQ (zmm)"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (EVEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "tremont": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "aes",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "fsgsbase",
+     "fxsr",
+     "gfni",
+     "mmx",
+     "movbe",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "prfchw",
+     "ptwrite",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sha",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "westmere": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "wildcatlake": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avxifma",
+     "avxneconvert",
+     "avxvnni",
+     "avxvnniint16",
+     "avxvnniint8",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "clwb",
+     "cmov",
+     "cmpccxadd",
+     "crc32",
+     "cx16",
+     "cx8",
+     "enqcmd",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "gfni",
+     "hreset",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "movdir64b",
+     "movdiri",
+     "nopl",
+     "pclmul",
+     "pconfig",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "ptwrite",
+     "rdpid",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "serialize",
+     "sha",
+     "sha512",
+     "shstk",
+     "sm3",
+     "sm4",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "uintr",
+     "vaes",
+     "vpclmulqdq",
+     "waitpkg",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (VEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "x86-64": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "cmov",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "sse",
+     "sse2",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "x86-64-v2": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "64bit",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "popcnt",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "x86-64-v3": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "avx",
+     "avx2",
+     "bmi",
+     "bmi2",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "f16c",
+     "fma",
+     "fxsr",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "popcnt",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "x86-64-v4": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "avx",
+     "avx2",
+     "avx512bw",
+     "avx512cd",
+     "avx512dq",
+     "avx512f",
+     "avx512vl",
+     "bmi",
+     "bmi2",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "f16c",
+     "fma",
+     "fxsr",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "nopl",
+     "popcnt",
+     "sahf",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "ssse3",
+     "x87",
+     "xsave"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "yonah": {
+    "backend": "simd_avx2 (no avx2: array polyfill)",
+    "features": [
+     "cmov",
+     "cx8",
+     "fxsr",
+     "mmx",
+     "nopl",
+     "sse",
+     "sse2",
+     "sse3",
+     "x87"
+    ],
+    "primitives": {
+     "aes_round": [],
+     "clmul": [],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)"
+     ],
+     "permute_var_u32": [],
+     "shift_var_u32": []
+    }
+   },
+   "znver1": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "clzero",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "mwaitx",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "prfchw",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sha",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "sse4a",
+     "ssse3",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "znver2": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "clwb",
+     "clzero",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fxsr",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "mwaitx",
+     "nopl",
+     "pclmul",
+     "popcnt",
+     "prfchw",
+     "rdpid",
+     "rdpru",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sha",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "sse4a",
+     "ssse3",
+     "wbnoinvd",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "znver3": {
+    "backend": "simd_avx2",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "clwb",
+     "clzero",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fsrm",
+     "fxsr",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "mwaitx",
+     "nopl",
+     "pclmul",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "rdpid",
+     "rdpru",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sha",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "sse4a",
+     "ssse3",
+     "vaes",
+     "vpclmulqdq",
+     "wbnoinvd",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC"
+     ],
+     "clmul": [
+      "PCLMULQDQ"
+     ],
+     "ifma52": [],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "znver4": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avx512bf16",
+     "avx512bitalg",
+     "avx512bw",
+     "avx512cd",
+     "avx512dq",
+     "avx512f",
+     "avx512ifma",
+     "avx512vbmi",
+     "avx512vbmi2",
+     "avx512vl",
+     "avx512vnni",
+     "avx512vpopcntdq",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "clwb",
+     "clzero",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fsrm",
+     "fxsr",
+     "gfni",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "mwaitx",
+     "nopl",
+     "pclmul",
+     "pku",
+     "popcnt",
+     "prfchw",
+     "rdpid",
+     "rdpru",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sha",
+     "shstk",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "sse4a",
+     "ssse3",
+     "vaes",
+     "vpclmulqdq",
+     "wbnoinvd",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC",
+      "VAESENC (zmm)"
+     ],
+     "clmul": [
+      "PCLMULQDQ",
+      "VPCLMULQDQ (zmm)"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (EVEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   },
+   "znver5": {
+    "backend": "simd_avx512",
+    "features": [
+     "64bit",
+     "adx",
+     "aes",
+     "avx",
+     "avx2",
+     "avx512bf16",
+     "avx512bitalg",
+     "avx512bw",
+     "avx512cd",
+     "avx512dq",
+     "avx512f",
+     "avx512ifma",
+     "avx512vbmi",
+     "avx512vbmi2",
+     "avx512vl",
+     "avx512vnni",
+     "avx512vp2intersect",
+     "avx512vpopcntdq",
+     "avxvnni",
+     "bmi",
+     "bmi2",
+     "clflushopt",
+     "clwb",
+     "clzero",
+     "cmov",
+     "crc32",
+     "cx16",
+     "cx8",
+     "f16c",
+     "fma",
+     "fsgsbase",
+     "fsrm",
+     "fxsr",
+     "gfni",
+     "invpcid",
+     "lzcnt",
+     "mmx",
+     "movbe",
+     "movdir64b",
+     "movdiri",
+     "mwaitx",
+     "nopl",
+     "pclmul",
+     "pku",
+     "popcnt",
+     "prefetchi",
+     "prfchw",
+     "rdpid",
+     "rdpru",
+     "rdrnd",
+     "rdseed",
+     "sahf",
+     "sha",
+     "shstk",
+     "sse",
+     "sse2",
+     "sse3",
+     "sse4.1",
+     "sse4.2",
+     "sse4a",
+     "ssse3",
+     "vaes",
+     "vpclmulqdq",
+     "wbnoinvd",
+     "x87",
+     "xsave",
+     "xsavec",
+     "xsaveopt",
+     "xsaves"
+    ],
+    "primitives": {
+     "aes_round": [
+      "AESENC",
+      "VAESENC (zmm)"
+     ],
+     "clmul": [
+      "PCLMULQDQ",
+      "VPCLMULQDQ (zmm)"
+     ],
+     "ifma52": [
+      "VPMADD52LUQ/HUQ (EVEX)"
+     ],
+     "mul_lo32": [
+      "PMULUDQ (xmm, SSE2)",
+      "VPMULUDQ (ymm, AVX2)",
+      "VPMULUDQ (zmm, AVX-512F)"
+     ],
+     "permute_var_u32": [
+      "VPERMD"
+     ],
+     "shift_var_u32": [
+      "VPSLLVD / VPSRLVD"
+     ]
+    }
+   }
+  }
+ },
+ "llvm_tag": "llvmorg-22.1.8"
+}