Po Yen Chen 95e722a3b3 [CK_TILE] Optimize fmha splitkv & splitkv combine kernels (#1577) ...

* Use smaller width for lse_accum dist tensor

* Update pipeline comment

* Fix wrong distribution for lse_accum

* Remove duplicate dim in lse_accum dist encoding

* Decide fmha splitkv combine kernel kBlockSize by kM0

* Remove assumption of MPerThread=1

* Add log<4> & log<8> specialization

* Enlarge occupancy array

* Fix vector size for small tile

* Add support for kMaxSplits=8

* Re-format gemm.hpp

* Use 16x16x16 warp gemm for fwd_splitkv

* Centralize policy code changes

* Leave fp8/bf8 tile settings unchanged

2024-10-21 10:52:11 +08:00

..

ops

[CK_TILE] Optimize fmha splitkv & splitkv combine kernels (#1577)

2024-10-21 10:52:11 +08:00

__init__.py

[CK_TILE] fmha forward split-kv + combine kernels (#1338)

2024-06-26 17:41:15 +08:00

cmake_config.py

[CK_TILE] fmha forward split-kv + combine kernels (#1338)

2024-06-26 17:41:15 +08:00

cpp_symbol_map.py

[CK_TILE] Add PagedAttention kernels (#1387)

2024-08-28 20:50:43 +08:00