Mesa (master): radeonsi: compute perf tests - don't test 1 wave/SA limit, test no limit first
GitLab Mirror
gitlab-mirror at kemper.freedesktop.org
Fri May 15 22:24:22 UTC 2020
Module: Mesa
Branch: master
Commit: 5acf99e81f7ad3c93426e62d6df98d64d73d80b3
URL: http://cgit.freedesktop.org/mesa/mesa/commit/?id=5acf99e81f7ad3c93426e62d6df98d64d73d80b3
Author: Marek Olšák <marek.olsak at amd.com>
Date: Wed May 6 19:32:01 2020 -0400
radeonsi: compute perf tests - don't test 1 wave/SA limit, test no limit first
1 wave/SA is always slow and thus not useful
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer at amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4935>
---
src/gallium/drivers/radeonsi/si_test_dma_perf.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/gallium/drivers/radeonsi/si_test_dma_perf.c b/src/gallium/drivers/radeonsi/si_test_dma_perf.c
index 116bfe69069..1613351611c 100644
--- a/src/gallium/drivers/radeonsi/si_test_dma_perf.c
+++ b/src/gallium/drivers/radeonsi/si_test_dma_perf.c
@@ -45,7 +45,7 @@ void si_test_dma_perf(struct si_screen *sscreen)
struct si_context *sctx = (struct si_context *)ctx;
const uint32_t clear_value = 0x12345678;
static const unsigned cs_dwords_per_thread_list[] = {64, 32, 16, 8, 4, 2, 1};
- static const unsigned cs_waves_per_sh_list[] = {1, 2, 4, 8, 16, 0};
+ static const unsigned cs_waves_per_sh_list[] = {0, 2, 4, 8, 16};
#define NUM_SHADERS ARRAY_SIZE(cs_dwords_per_thread_list)
#define NUM_METHODS (4 + 2 * NUM_SHADERS * ARRAY_SIZE(cs_waves_per_sh_list))
More information about the mesa-commit
mailing list