ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-11 08:50:17 +00:00
Code Issues Packages Projects Releases Wiki Activity

Updating Runtime log for CK Tile Engine (#2431)

Browse Source 
* Updating runtime log message for CK TILE ENGINE

* Fixing Clang Format

* Update tile_engine/ops/gemm/README.md

Co-authored-by: Aviral Goel <aviral.goel@amd.com>

---------

Co-authored-by: ThruptiRajLakshmanaGowda <tlakshma@amd.com>
Co-authored-by: Aviral Goel <aviral.goel@amd.com>
This commit is contained in:
Thrupti Raj Lakshmana Gowda
2025-07-01 12:59:49 -05:00
committed by GitHub
parent e9036a8fc2
commit a03682cb80
3 changed files with 9 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
tile_engine/ops/gemm/README.md
View File

@@ -4,7 +4,8 @@ CK Tile Engine GEMM is used to generate and run GEMM kernels with different comb
# Kernel Configurations
User can provide kernel configuration such as tile size, warp size, padding, pipeline, scheduler and epilogue in the config file with limited values. For reference please see `./configs/user_provided_config.json`.
Users can specify custom kernel configurations such as tile size, warp size, padding, pipeline, scheduler, and epilogue in the config file. This allows building only for selected configurations, significantly reducing build time.
For reference please see `./configs/user_provided_config.json`.
The Tile engine also has a default kernel configuration for providing range of configuration parameter values, which helps users who lack kernel development experience to benchmark. For reference please see in `./configs/default_config.json`

8
tile_engine/ops/gemm/benchmark_gemm.hpp
View File

@@ -163,7 +163,8 @@ auto calculate_rtol_atol(const ck_tile::index_t K,
}
/// @brief Function to compare the results of the device and host computations
bool compare(ck_tile::index_t K,
bool compare(std::string instanceName,
ck_tile::index_t K,
ck_tile::index_t kbatch,
ck_tile::HostTensor<CDataType>& c_m_n_dev_result,
ck_tile::HostTensor<CDataType>& c_m_n_host_result)
@@ -178,8 +179,9 @@ bool compare(ck_tile::index_t K,
rtol_atol.at(ck_tile::number<0>{}),
rtol_atol.at(ck_tile::number<1>{}));
std::cout << "Relative error threshold: " << rtol_atol.at(ck_tile::number<0>{})
<< " Absolute error threshold: " << rtol_atol.at(ck_tile::number<1>{}) << std::endl;
std::cout << "For " << instanceName << " Relative error threshold is "
<< rtol_atol.at(ck_tile::number<0>{}) << " Absolute error threshold is "
<< rtol_atol.at(ck_tile::number<1>{}) << std::endl;
std::cout << "The verification result is:" << (pass ? "correct" : "fail") << std::endl;
return pass;

3
tile_engine/ops/gemm/gemm_profiler.hpp
View File

@@ -172,7 +172,8 @@ class GemmProfiler
c_m_n_dev_buf.FromDevice(c_m_n_dev_result.data());
bool verified_correct =
!setting_.verify_ ||
compare(gemm_problem.k_, gemm_problem.split_k_, c_m_n_dev_result, c_m_n_host_result);
compare(
name, gemm_problem.k_, gemm_problem.split_k_, c_m_n_dev_result, c_m_n_host_result);
if(verified_correct)
{