github/ROCm
1
1
Fork 0
mirror of https://github.com/ROCm/ROCm.git synced 2026-04-05 03:01:17 -04:00
Code Issues Packages Projects Releases Wiki Activity

[TESTS] Increase the paylog of the globaltimer kernel to reduce chances of fakiness (#1726)

Browse Source 
If the kernel is too small, on a very fast GPU we may get 0 because the
resolution of the timer seems too coarse.

Fixes #1725
This commit is contained in:
Mehdi Amini
2023-06-01 02:53:07 -07:00
committed by GitHub
parent b0c893cdc5
commit 440fd1bf20
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
python/test/unit/language/test_core.py
View File

@@ -2852,7 +2852,7 @@ def test_globaltimer():
def kernel(Out1, Out2):
start = tl.extra.cuda.globaltimer()
off = tl.arange(0, 128)
for i in range(100):
for i in range(10000):
tl.store(Out1 + off, tl.load(Out1 + off) + 1)
end = tl.extra.cuda.globaltimer()
tl.store(Out2, end - start)