kfdtest: decrease granularityMB for handling small vram system
It's not possible to allocate the 3/4 vram size with granularityMB being 128 when vram size < 512MB and decrease granularityMB to 16 has no significant impact on ROCt test on other system. So let's decrease granularityMB on small vram system for handling LargestVramBufferTest(). Change-Id: Iea7c29abfd382a20761b653730fd09a220ad2fd0 Signed-off-by: Prike Liang <Prike.Liang@amd.com>
Этот коммит содержится в:
@@ -965,7 +965,7 @@ void KFDMemoryTest::BinarySearchLargestBuffer(int allocNode, const HsaMemFlags &
|
||||
int ret;
|
||||
|
||||
HsaMemMapFlags mapFlags = {0};
|
||||
HSAuint64 granularityMB = 128;
|
||||
HSAuint64 granularityMB = highMB > 512 ? 128 : 16;
|
||||
|
||||
/* Testing big buffers in VRAM */
|
||||
unsigned int * pDb = NULL;
|
||||
|
||||
Ссылка в новой задаче
Block a user