Po Yen Chen 54f0e6f4bb [CK_TILE] More fmha splitkv optimizations (#1588) ...

* Use pre-defined constants for readability

* Use vector write for o_acc tensor

* Remove no-longer used policy method

* Deprecate no-longer used policy/pipeline

* Specify gemm0/gemm1 block warps separately in codegen

* Fix wrong ps_idx creation logic

* Add single-warp block gemm

* Supoprt single-warp gemm0

* Make MakeCBlockTile() as static method

* Use MakeCBlockTile() to get underlying tile distribution

* Use kNumGemm1Warps to compute # threads for gemm1

* Put normal case in the if clause

* Refine fmha splitkv block mapping

* Refine & fix the lse_acc/o_acc layout

* Fix wrong LDS size for K tile

* Use kK0=64 for hdim=128,256 fmha splitkv kernels

* Use kK1=64 for hdim=32,64,128 fmha splitkv kernels

* Undo kK0/kK1 changes

* Use more reasonable GetAlignmentV() computation

* Using store_tile() in fmha splitkv kernel epilogue

2024-10-26 18:35:45 +08:00

..

01_fmha

[CK_TILE] More fmha splitkv optimizations (#1588)

2024-10-26 18:35:45 +08:00

02_layernorm2d

update layernorm (#1570)

2024-10-22 09:26:18 +08:00

03_gemm

[CK_TILE] Add block universal gemm pipeline policy (#1557)

2024-10-15 13:53:41 +02:00

04_img2col

[CK_TILE] Update example README files & fix script compatibility issue (#1548)

2024-10-08 10:45:12 +08:00

05_reduce

update layernorm (#1570)

2024-10-22 09:26:18 +08:00

CMakeLists.txt

update layernorm (#1570)

2024-10-22 09:26:18 +08:00

remod.py

introducing ck_tile! (#1216)

2024-04-15 19:27:12 -05:00