[github-actions]

This pull request changes the compat entry for the Finch package from 1.2.9 to 1.2.9, 2.
This keeps the compat entries for earlier versions.

Note: I have not tested your package with this new compat entry.
It is your responsibility to make sure that your package tests pass before you merge this pull request.

[github-actions]

Dagger benchmarks: dirty vs master

Counts are benchmark metrics (time, allocations, bytes). Timing changes require non-overlapping median ± IQR bands; both revisions need at least five timed samples, and timing regressions need an independent confirmation run.

Backend	Regressions	Improvements	Within noise	Inconclusive time	Suites
Threads	1	1	16	0	4/4
Distributed	8	0	2	0	4/4
Distributed+Threads	0	0	1	0	4/4
MPI	0	0	0	0	4/4
MPI+Threads	1	0	13	1	4/4
Total	10	1	32	1	20/20

Regressions across all backends

Backend	Benchmark	Metric	Change
Distributed	linalg/dagger/N=1024 (block 128)/matmul (A*A)	memory	+115.5%
Distributed	linalg/dagger/N=1024 (block 128)/matmul (A*A)	allocs	+109.8%
Distributed	linalg/dagger/N=1024 (block 128)/lu	memory	+74.2%
MPI+Threads	linalg/dagger/N=1024 (block 512)/solve (A\b via lu)	time	+70.9%
Distributed	linalg/dagger/N=1024 (block 128)/lu	allocs	+65.8%
Distributed	linalg/dagger/N=1024 (block 128)/solve (A\b via lu)	memory	+61.7%
Distributed	linalg/dagger/N=1024 (block 128)/solve (A\b via lu)	allocs	+51.5%
Threads	linalg/dagger/N=1024 (block 128)/svd	allocs	+34.3%
Distributed	sparse/dagger/N=1024 (block 64)/cg solve (laplacian)	memory	+31.8%
Distributed	sparse/dagger/N=1024 (block 64)/cg solve (laplacian)	allocs	+30.8%

Improvements across all backends

Backend	Benchmark	Metric	Change
Threads	stencil/dagger/N=256 (block 128)/neighbors (Reflect)	time	-62.4%

Threads (1 process × 4 threads) — 1 regressions, 1 improvements

Suite	Regressions	Improvements	Within noise	Inconclusive time
array	0	0	8	0
linalg	1	0	2	0
sparse	0	0	0	0
stencil	0	1	6	0

Regressions

Benchmark	Metric	Change
linalg/dagger/N=1024 (block 128)/svd	allocs	+34.3%

Improvements

Benchmark	Metric	Change
stencil/dagger/N=256 (block 128)/neighbors (Reflect)	time	-62.4%

Within noise

Benchmark	Metric	Change
array/dagger/N=1024 (block 512)/broadcast (X .+ 1)	time	+138.3%
array/dagger/N=1024 (block 512)/norm	time	+136.7%
array/dagger/N=256 (block 128)/add (X + X)	time	+108.5%
array/dagger/N=1024 (block 512)/reduce (sum)	time	+72.3%
array/dagger/N=256 (block 128)/broadcast (X .+ 1)	time	+69.5%
stencil/dagger/N=1024 (block 128)/assign (const)	time	+45.7%
array/dagger/N=256 (block 128)/reduce (sum)	time	+34.2%
linalg/dagger/N=1024 (block 512)/matvec (A*x)	time	+33.9%
array/dagger/N=256 (block 256)/norm	time	+27.0%
array/dagger/N=256 (block 256)/broadcast (X .+ 1)	time	+26.0%
stencil/dagger/N=256 (block 128)/neighbors (Clamp)	time	+25.2%
stencil/dagger/N=1024 (block 512)/neighbors (Clamp)	time	-28.2%
stencil/dagger/N=1024 (block 512)/multi-expr	time	-29.9%
stencil/dagger/N=256 (block 128)/multi-expr	time	-42.6%
linalg/dagger/N=256 (block 128)/syrk (A'*A)	time	-43.2%
stencil/dagger/N=256 (block 128)/update (+)	time	-53.4%

Distributed (4 processes × 1 thread) — 8 regressions, 0 improvements

Suite	Regressions	Improvements	Within noise	Inconclusive time
array	0	0	0	0
linalg	6	0	2	0
sparse	2	0	0	0
stencil	0	0	0	0

Regressions

Benchmark	Metric	Change
linalg/dagger/N=1024 (block 128)/matmul (A*A)	memory	+115.5%
linalg/dagger/N=1024 (block 128)/matmul (A*A)	allocs	+109.8%
linalg/dagger/N=1024 (block 128)/lu	memory	+74.2%
linalg/dagger/N=1024 (block 128)/lu	allocs	+65.8%
linalg/dagger/N=1024 (block 128)/solve (A\b via lu)	memory	+61.7%
linalg/dagger/N=1024 (block 128)/solve (A\b via lu)	allocs	+51.5%
sparse/dagger/N=1024 (block 64)/cg solve (laplacian)	memory	+31.8%
sparse/dagger/N=1024 (block 64)/cg solve (laplacian)	allocs	+30.8%

Within noise

Benchmark	Metric	Change
linalg/dagger/N=1024 (block 128)/matmul (A*A)	time	+59.9%
linalg/dagger/N=1024 (block 128)/syrk (A'*A)	time	+44.6%

Distributed+Threads (2 processes × 2 threads) — 0 regressions, 0 improvements

Suite	Regressions	Improvements	Within noise	Inconclusive time
array	0	0	1	0
linalg	0	0	0	0
sparse	0	0	0	0
stencil	0	0	0	0

Within noise

Benchmark	Metric	Change
array/dagger/N=1024 (block 128)/broadcast (X .+ 1)	time	+36.7%

MPI (4 ranks × 1 thread) — 0 regressions, 0 improvements

Suite	Regressions	Improvements	Within noise	Inconclusive time
array	0	0	0	0
linalg	0	0	0	0
sparse	0	0	0	0
stencil	0	0	0	0

MPI+Threads (2 ranks × 2 threads) — 1 regressions, 0 improvements

Suite	Regressions	Improvements	Within noise	Inconclusive time
array	0	0	7	1
linalg	1	0	2	0
sparse	0	0	0	0
stencil	0	0	4	0

Regressions

Benchmark	Metric	Change
linalg/dagger/N=1024 (block 512)/solve (A\b via lu)	time	+70.9%

Within noise

Benchmark	Metric	Change
array/dagger/N=256 (block 256)/reduce (sum)	time	+181.2%
stencil/dagger/N=1024 (block 512)/assign (const)	time	+82.8%
array/dagger/N=256 (block 128)/add (X + X)	time	+82.8%
array/dagger/N=1024 (block 512)/transpose (permutedims)	time	+79.5%
array/dagger/N=1024 (block 512)/alloc (rand)	time	+68.4%
linalg/dagger/N=256 (block 256)/cholesky	time	+51.2%
stencil/dagger/N=1024 (block 512)/neighbors (Wrap)	time	+46.8%
linalg/dagger/N=1024 (block 512)/cholesky	time	+46.1%
array/dagger/N=1024 (block 128)/norm	time	+37.0%
stencil/dagger/N=256 (block 128)/multi-expr	time	-36.0%
stencil/dagger/N=1024 (block 512)/multi-expr	time	-44.3%
array/dagger/N=256 (block 128)/alloc (rand)	time	-55.9%
array/dagger/N=1024 (block 512)/reduce (sum)	time	-60.8%

Inconclusive timing changes

Benchmark	Metric	Change
array/dagger/N=256 (block 128)/reduce (sum)	time	+398.1%

Full results and plots (download the benchmark-results-* artifacts).