ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-04-19 22:39:03 +00:00
Code Issues Packages Projects Releases Wiki Activity

[CK][EXAMPLES] (#2826)

Browse Source 
-Added parameter to enable/disable verification and timing of kernel in various examples that missed it.
-Added parameter to change number of groups to execute in grouped_gemm_examples.

Signed-off-by: Michal Kulikowski <Michal.Kulikowski@amd.com>
This commit is contained in:
Michał Kulikowski
2025-09-11 21:33:00 +02:00
committed by GitHub
parent f3239395dc
commit ffe9775e70
33 changed files with 578 additions and 116 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
example/44_elementwise_permute/elementwise_trinary_4D_fp16.cpp
View File

@@ -48,11 +48,27 @@ using DeviceElementwisePermuteInstance = ck::tensor_operation::device::DeviceEle
ck::Sequence<8, 8, 8>, // InScalarPerVectorSeq
ck::Sequence<8>>; // OutScalarPerVectorSeq
int main()
int main(int argc, char* argv[])
{
bool do_verification = true;
bool time_kernel = true;
if(argc == 1)
{
// use default
}
else if(argc == 3)
{
do_verification = std::stoi(argv[1]);
time_kernel = std::stoi(argv[2]);
}
else
{
printf("arg1: verification (0=no, 1=yes)\n");
printf("arg2: time kernel (0=no, 1=yes)\n");
exit(0);
}
std::vector<std::size_t> nchw = {16, 128, 32, 64};
std::array<ck::index_t, 4> ab_lengths;
std::array<ck::index_t, 4> ab_strides = {static_cast<int>(nchw[1] * nchw[2] * nchw[3]),