mirror of
https://github.com/ROCm/composable_kernel.git
synced 2026-04-20 06:49:15 +00:00
CK-Tile first draft of universal block gemm with interwave & intrawave scheduler (#1676)
* Block universal gemm. * Universal block gemm with interwave scheduler - draft. * Refactoring * Move a/b_warp_tiles into BlockGemmImpl * set BlockGemmImpl as a class member * Change tile size for more suitable to memory bound cases. * Introduce kKPerThread to WarpGemm * Add documentation comment. * Fix Interwave scheduler block gemm. * Add compute/memory friendly tile configuration. * Clean * New tile configurations in gemm mem example. * Add more static checks and fix loop order in block gemm. * Add more static checks and use warp gemm mfma dispatcher. * Add default scheduler block gemm. * Remove logging in example.
This commit is contained in:
@@ -261,7 +261,7 @@ bool run_gemm(const ProblemType& problem_size, const ExecutionConfig& config)
|
||||
if(config.time_kernel)
|
||||
{
|
||||
ave_time =
|
||||
invoker.Run(argument, StreamConfig{nullptr, config.time_kernel, 0, 5, 10, true, 4});
|
||||
invoker.Run(argument, StreamConfig{nullptr, config.time_kernel, 0, 50, 100, true, 4});
|
||||
|
||||
std::size_t flop = 2_uz * M * N * K;
|
||||
std::size_t num_btype =
|
||||
|
||||
Reference in New Issue
Block a user