ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-06-08 15:30:23 +00:00
Code Issues Packages Projects Releases Wiki Activity

Add persistent async input scheduler for GEMM kernels (#3520)

Browse Source 
Add signal-based synchronization for persistent GEMM kernels where
input data becomes available incrementally. Uses modulo wraparound
(like PyTorch's AsyncMM) for chunk index calculation:
  chunk_idx = ((tile_idx + tile_idx_pivot) / tiles_per_chunk) % num_chunks

Key components:
- PersistentAsyncInputScheduler struct with tiles_per_chunk_m,
  chunk_signals, tile_idx_pivot_m, and num_chunks fields
- wait_eq_wave method using __builtin_amdgcn_s_sleep for power efficiency
- IsSupportedArgument validation for scheduler parameters
- Example demonstrating async input scheduling with simulated producer
- GTest unit tests covering all layout combinations
This commit is contained in:
Max Podkorytov
2026-01-20 10:37:09 -08:00
committed by GitHub
parent 8f75869408
commit 91b4102a59
11 changed files with 844 additions and 61 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
include/ck_tile/core.hpp
View File

@@ -91,6 +91,7 @@
#include "ck_tile/core/utility/ignore.hpp"
#include "ck_tile/core/utility/literals.hpp"
#include "ck_tile/core/utility/magic_div.hpp"
#include "ck_tile/core/utility/persistent_async_input_scheduler.hpp"
#include "ck_tile/core/utility/philox_rand.hpp"
#include "ck_tile/core/utility/print.hpp"
#include "ck_tile/core/utility/random.hpp"