mirror of
https://github.com/ROCm/composable_kernel.git
synced 2026-05-11 08:50:17 +00:00
Input/output permutation for fused attention (#460)
* reopen masking att instance due to CI is upgraded * re-enable instances previously failed on 9110 * enable ksize-kpadding pair validity test * add non-masked attention+permute test; expose masking boolean to attention kernel handles * disable bench * fix test * move files * bulk rename batched_gemm_masking_scale_softmax_gemm_permute to batched_gemm_softmax_gemm_permute * format * amend rename * disable bench in test * add mask/no-mask test for non-permute attention kernels * disable broken kernel instance * example working add non-permuted problem statement evaluating whether overhead comes from permutation or the extra kernel arg * interface for bias addition without implementing it * test and profiler running * tidy * mask type determined by enum class * unify example code * move masking specialization to its own header * align formats * extract helper functions * experiment merging dims for attn w/ permute; shows perf parity with attn wo/ permute * add tensor specialization to template args since tensor spec packed shows perf parity when permutation isn't needed remove redundant template args comment on 'packed' tensor specialization * grouped attention with input/output permute example * format * clean up * refactor acc0 tile visitor Co-authored-by: shaojiewang <wsjmessi@163.com> Co-authored-by: Chao Liu <chao.liu2@amd.com>
This commit is contained in:
@@ -159,6 +159,11 @@
|
||||
// tuning parameter
|
||||
#define CK_WORKAROUND_SWDEV_325164 0
|
||||
|
||||
// workaround: disable broken fused attention kernel instance that does not pass validation
|
||||
// issue found on mi100/#10738 combo when irregular KPerBlock attention kernel has acc0 scaling
|
||||
// enabled
|
||||
#define CK_WORKAROUND_DISABLE_BROKEN_ATTN_KERNEL_INSTANCE 1
|
||||
|
||||
namespace ck {
|
||||
|
||||
enum struct InMemoryDataOperationEnum
|
||||
|
||||
Reference in New Issue
Block a user