From 85d67eeffb7bc51a9c04485aa7eb695162dcf986 Mon Sep 17 00:00:00 2001
From: All Asmaul Husnain <2263081025@uttarauniversity.edu.bd>
Date: Sat, 10 Oct 2026 17:28:35 +0600
Subject: [PATCH] Clarify benchmark timing and measured work

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index 5433957..53b1b98 100644
--- a/README.md
+++ b/README.md
@@ -141,7 +141,7 @@ B tree pages use a compact fixed layout designed to keep the storage implementat
 
 ## Performance
 
-The benchmark inserts 1,000 rows through the public table API and measures the insertion phase. It is intended for repeatable local comparisons rather than as a claim of production database performance.
+The benchmark inserts 1,000 rows through the public table API, then flushes the database pages. It reports process CPU time from `clock()`, including both the insert loop and the flush but excluding database opening. It is intended for repeatable local comparisons rather than as a claim of production database performance.
 
 The current write path keeps modified pages in the pager cache during inserts and persists them during an explicit flush. This avoids performing a file write and fflush for every inserted row.