23 lines
891 B
Diff
23 lines
891 B
Diff
|
From 8a1710dd0da445d76e6eaeb35b180d24efac0919 Mon Sep 17 00:00:00 2001
|
||
|
|
From: Martin Kroeker <martin@ruby.chemie.uni-freiburg.de>
|
||
|
|
Date: Sun, 6 Oct 2024 20:03:32 +0200
|
||
|
|
Subject: [PATCH] don't apply switch_ratio to tail of loop
|
||
|
|
|
||
|
|
---
|
||
|
|
driver/level3/level3_thread.c | 2 +-
|
||
|
|
1 file changed, 1 insertion(+), 1 deletion(-)
|
||
|
|
|
||
|
|
diff --git a/driver/level3/level3_thread.c b/driver/level3/level3_thread.c
|
||
|
|
index ddb39abd66..3d56c45a99 100644
|
||
|
|
--- a/driver/level3/level3_thread.c
|
||
|
|
+++ b/driver/level3/level3_thread.c
|
||
|
|
@@ -742,7 +742,7 @@ static int gemm_driver(blas_arg_t *args, BLASLONG *range_m, BLASLONG
|
||
|
|
num_parts = 0;
|
||
|
|
while (n > 0){
|
||
|
|
width = blas_quickdivide(n + nthreads - num_parts - 1, nthreads - num_parts);
|
||
|
|
- if (width < switch_ratio) {
|
||
|
|
+ if (width < switch_ratio && width > 1) {
|
||
|
|
width = switch_ratio;
|
||
|
|
}
|
||
|
|
width = round_up(n, width, GEMM_PREFERED_SIZE);
|