ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-06-30 19:57:40 +00:00
Code Issues Packages Projects Releases Wiki Activity
1,914 Commits 776 Branches 40 Tags
ff4d6434d96b0443ddc4aee59acbc074edeb4ae2
Commit Graph

222 Commits

Author SHA1 Message Date
Feng Shijie
ff4d6434d9 optimize grid calculation in contiguous mode 2025-07-09 08:24:16 +00:00
Feng Shijie
fae4ebac66 Add support for contiguous grouped gemm 2025-07-09 07:03:30 +00:00
Feng Shijie
0deeba90e6 prune debug code 2025-07-04 08:06:43 +00:00
Feng Shijie
bce9c22bcd Improve gridDim calculation in persistent mode 2025-07-04 07:09:58 +00:00
Feng Shijie
10fe5ab7b5 Add group-flatmm 2025-07-04 03:37:43 +00:00
root
56f84349ca 1.fix loop_num=odd bug 2.optimize mi300 performance of big MNK(tilesize 128x128x128) 3.optimize decode perf on mi300 2025-06-25 20:41:25 -05:00
root
40f1d5829e add define for different wave tile 2025-06-19 02:20:40 -05:00
root
a6f711de98 support even odd k in tail 2025-06-17 02:31:54 -05:00
root
bd347831e0 compile IGLP work aroung 2025-05-30 00:58:56 -05:00
root
409cd1fa72 enable hotloop 2025-05-28 02:12:13 -05:00
root
34a8c08dc8 re-org tail code 2025-05-28 01:34:24 -05:00
root
0ef8ad46c0 1.control main loop pipeline by C code instead of hotloop. 2.seperate 2 lds bank 2025-05-26 09:48:58 -05:00
root
c097f90adf first version 2025-05-26 04:45:18 -05:00
solin
c2d17ec24f draft for 16*16*128 fp8 2025-05-14 09:56:13 +00:00
BingYuan.Zhou
41c17d0a95 fix moe sorting build fail (#2190) 
* fix moe sorting build fail

* refile code

---------

Co-authored-by: solin <bingzhou@amd.com>
 2025-05-14 09:31:26 +08:00
Illia Silin
58f9e9ffbc Update the buffer load/store intrinsic names for clang>=20. (#2192) 
* fix the buffer load/store intrinsic names

* fix clang format
 2025-05-13 10:18:14 -07:00
Po Yen Chen
2920604786 [CK_TILE] Add logits soft-capping & customization support to the FMHA forward kernel/pipelines (#2163) 
* hack for cap logits

* fix bug

* Re-format files

* Allow specifying logits_soft_cap through APIs

* Support turn on/off logits_soft_cap in async pipeline

* Do not generate non-verified kernels

* Align receipt used in Aiter

* Sync logits soft-capping across pipelines

* Re-enable some hdim pipelines

* fix perf

* Add attention variant for logits_soft_cap

* Add newline at end-of-file

* Fix performance

* Add comment to explain logits_soft_cap pre-processing

* Unify code

* Unify floating-point literal style

* Use class data member to slience the compilation error

* [CK_TILE] Update attention customizaton interface: add LogitsMask() (#2133)

* Send 'mask' along with variant params to the LogitsMask()

* Send block indices to the variant

* Add indices parameters in variant interface

* Fix fmha bwd codegen error

* Allow switch logits_soft_cap impl

* Eliminate register spills

* Fix compilation errors

* Fix wrong LSE

* Fix LSE for splitkv kernel

* Sync splitkv pipeline changes

* Add batch_prefill kernel/pipeline

* Fix codegen error

* Undo changes in CMakeLists.txt

* Merge pipeline filtering check

* Use different code path if kHasLogitsSoftCap=false

* Remove [[maybe_unused]] attribute

* Use pre-existing compile-time flag to instantiate templates

* Sync pipeline changes

* Update CHANGELOG.md

---------

Co-authored-by: Bernard <bernaliu@amd.com>
Co-authored-by: coderfeli <coderfeli@163.com>
 2025-05-13 12:19:25 +08:00
Khushbu Agarwal
f05e45ba59 Disable SMFMA gfx90a (#2184) 
* sparsity fix for gfx90a

* reverting tile_engine changes
 2025-05-12 09:56:23 -07:00
Thomas Ning
9d1e44e56a Vectorized Transpose for Batched Transpose CK Tile Operator (#2131) 
* Shared Memory for single data point

* CKTile Transpose vectorize CP1

* CKTile Transpose vectorize CP2

* CKTile Transpose vectorize CP2.1

* fixed the compile error of the transpose tile 2d

* Have the correct result for the current test sample

* Changes to printing tensor

* fp8 support added

* Debugging for transpose

* solving the corner issue

* Changed padding flag

* Intermideate Debugging

* Intermidiate Debugging

* Intermediate Debugging

* Finished debugging of the transpose op

* Code Cleanup

* Adding edge case smoke tests

* Adding Transpose test to CI/CD

* Adding Transpose test to CI/CD

* Adding Transpose test to CI/CD

* Addressing Review Comment

* Addressing Comments

* Addressing Comments

* Measuring Perf Tests

* Code Cleanup

* Changlog

* Added the running iterations

* clang format

* Fix the changelog

* Fix the compilation error

* change the printing factor

---------

Co-authored-by: ThruptiRajLakshmanaGowda <tlakshma@amd.com>
 2025-05-12 00:41:45 -07:00
Khushbu Agarwal
d8faf1c6a1 Support for swizzle and transpose for MFMA_16x16x32_F16/BF16 (#2172) 
* Changes for updating tile distribution for shuffle and transpose

* Fixed swizzle and transpose, removed comments

* clang formatted

* Adding support for bf16 type

* Addressing review comments
 2025-05-10 22:40:05 -07:00
Khushbu Agarwal
ef72a4b9bc Disable SMFMA for gfx90a (#2182) 2025-05-09 00:18:07 -07:00
Thomas Ning
c757046d49 Revert "Disable the SMFMA instruction for gfx90a. (#2174)" (#2175) 
This reverts commit a32d907771.
 2025-05-08 00:07:03 -07:00
Khushbu Agarwal
a32d907771 Disable the SMFMA instruction for gfx90a. (#2174) 
* remove smfma for gfx90a

* clang formatted
 2025-05-07 23:09:22 -07:00
BingYuan.Zhou
6a3960c1e1 Flatmm merge (#2168) 
* sync with function interface of cshuffleepiloge,fix flatmm build fail

* move code from solin/flatmm which add mfma16*16*32fp8 and optimize flatmm

---------

Co-authored-by: solin <bingzhou@amd.com>
 2025-05-08 12:59:57 +08:00
jakpiase
cb07ad84d5 fix for default epilogue (#2167) 2025-05-07 10:46:53 -07:00
Aviral Goel
769336b640 [CK_TILE] Add type traits to detect tile window types at compile time (#2158) 
* added WindowType enum to tile_window_structs and static assert checks in computev4 pipeline

* added type traits instead of enum to tile_window() and tile_window_linear() with debug comments

* removed comments, added documentation and clang format
 2025-05-07 00:00:39 -07:00
carlushuang
4e9b76f88c [CK_TILE] optimize moe sorting kernel, boost large context case up to 20x (#2153) 
* combine 2-3 as single stage

* support zeroing

* improve long tokens

* update specialization

* b16 ws

* 8bit topk optimize

* update 15 example
 2025-05-06 17:32:07 +08:00
jakpiase
0bcb804ad0 [CK_TILE] Remove scratch usage from universal gemm (#2001) 
* moves kbatch condition outside of kernel

* add reviewer comments

* fixes

* fix tests

* fixes after review

---------

Co-authored-by: Adam Osewski <19374865+aosewski@users.noreply.github.com>
 2025-05-05 18:46:44 +02:00
Khushbu Agarwal
d58f2b8bd0 mfma_32x32x64_fp8/bf8 (#2148) 
* support for mfma_32x32x64_fp8

* clang-formatted

* Fixing sparsity in codegen
 2025-05-01 13:36:24 -07:00
Aviral Goel
1d8ef40760 Add documentation for ck_tile::array<T,N> (#2078) 
* addded documentation for ck_tile::array<T,N>

* clang format fix

* spelling errros

Co-authored-by: Adam Osewski <19374865+aosewski@users.noreply.github.com>

* spelling errros

Co-authored-by: Adam Osewski <19374865+aosewski@users.noreply.github.com>

* Apply suggestions from code review

Co-authored-by: spolifroni-amd <Sandra.Polifroni@amd.com>

* Revert "spelling errros"

This reverts commit 4179e7d193.

* Revert "spelling errros"

This reverts commit 3f90733dbe.

---------

Co-authored-by: Adam Osewski <19374865+aosewski@users.noreply.github.com>
Co-authored-by: spolifroni-amd <Sandra.Polifroni@amd.com>
Co-authored-by: John Afaganis <john.afaganis@amd.com>
 2025-04-30 16:43:36 -07:00
Illia Silin
9a9f59ae69 Revert "Add ck tile examples to package (#1880)" (#2150) 2025-04-30 10:20:16 -07:00
Aviral Goel
65f182d617 Add Matrix A and Matrix B Swizzle for LDS in Computev4 policy (#2136) 
* fixed computev4 policy bug for lds swizzle

* added swizzle for input matrix B

* Improved ComputeV4 policy and pipeline by swizzling A and B

* consolidated LDS descriptor functions in parent struct
 2025-04-28 18:20:47 -07:00
Khushbu Agarwal
d107f3c3a5 Support for MFMA_16x16x128 for fp8/bf8 (#2125) 
* Adding 16x16x128 support for gfx950

* Support for fp8 and bf8

* fix input arguments for MFMA scale instruction

* clang-formatted

* Fixes for lwpck-3145 (#2138)

* Fix lds tile & cmake dep & default epilogue

* Fallback BTypeToUse to ADataType in WOQ cases

* reverting instance json file

* reverting instance json file

---------

Co-authored-by: Yi DING <yi.ding@amd.com>
 2025-04-28 18:19:50 -07:00
jakpiase
434d19f696 Add ck tile examples to package (#1880) 
* add ck tile examples to package

* Update jenkinsfile

* fix for jenkinsfile

* fix for building ck tile code on non gfx9

* compile ck tile examples only for gfx94

* include ck tile examples in all target

* fix for basic gemm UseStructuredSparsity

* Update CMakeLists.txt

* Update gemm_pipeline_problem.hpp

* add targets to rocm install

---------

Co-authored-by: Illia Silin <98187287+illsilin@users.noreply.github.com>
 2025-04-28 09:53:19 -07:00
Yi DING
8add2cf45d Fix fp8 convert & add option for basic example (#2129) 2025-04-27 16:26:05 -07:00
Khushbu Agarwal
a2ed34a112 MFMA_32x32x16 for gfx950 (#2121) 
* Enable MFMA_32x32x16 for fp16/BF16 for gfx950

* clang formatted
 2025-04-24 10:20:22 -07:00
Illia Silin
01cb8379cd make code compliant with std=c++20 (#2123) 2025-04-24 10:14:52 -07:00
carlushuang
5487289fc4 [CK_TILE] support gfx950 matrix core in 01_fmha fwd (#2110) 
* gfx950 01_fmha fwd

* fix comment

---------

Co-authored-by: Thomas Ning <Thomas.Ning@amd.com>
 2025-04-23 12:40:18 -07:00
Gino Lu
504f563f78 [CK-Tile] warp-gemm support for using V_MFMA_F32_16x16x32_BF16 (#2073) 
* draft v_mfma_f32_16x16x32_bf16

* fix error config and add debug code.

* Solve the CShuffle Problem

* draft v_mfma_f32_16x16x32_bf16

* fix error config and add debug code.

* Solve the CShuffle Problem

* fix error while testing new command

* Finished the feature of new mfma 16*16*32

* Addressed the comment

---------

Co-authored-by: ThomasNing <thomas.ning@amd.com>
 2025-04-22 15:52:36 -07:00
Thomas Ning
a738e43445 MFMA 16x16x32fp8 (#2103) 
* add mfma_16x16x32_fp8

* clang format code

* Finished the fix for gemm basic

* clang foramt

* rebuild CI

* recover gemm.hpp

* add MFMA 16*16*32bf8

---------

Co-authored-by: solin <bingzhou@amd.com>
 2025-04-21 10:21:35 -07:00
solin
c318ec0778 fix CI build fail 2025-04-21 16:00:12 +08:00
BingYuan.Zhou
eaf1f0bf3b [flatmm] implement basic fp16 flatmm (#2089) 
* [flatmm] implement basic fp16 flatmm

* fix CI build fail

---------

Co-authored-by: root <root@hjbog-srdc-50.amd.com>
Co-authored-by: solin <bingzhou@amd.com>
 2025-04-16 16:51:17 +08:00
Thomas Ning
269f4f6af5 Solve the Static Encoding Pattern compile error when the tile size is too small (#2079) 2025-04-13 20:09:30 -07:00
jakpiase
6c61f4d237 [CK_TILE] Add 2:4 structured sparsity support for fp16 gemm (#1957) 
* add structured sparsity fp16 support for gemm

* added reviewer suggestions

* update changelog

* update changelog

* add reviewers suggestions

* Minor fix

* clang fix

* fix doxygen
 2025-04-11 12:18:26 +02:00
slippedJim
5f885d2b7a add fmha fwd splitkv receipt for aiter c++ api (#2068) 
* add s_randval for c++ api

* Fix bug of bias in splitkv

---------

Co-authored-by: rocking <ChunYu.Lai@amd.com>
 2025-04-10 23:21:13 +08:00
Juan Manuel Martinez Caamaño
f14e648e7c Replace inline assembly with builtins in FHMA (#2067) 
* Replace inline assembly with builtins in FHMA

---------

Co-authored-by: illsilin <Illia.Silin@amd.com>
 2025-04-10 09:48:37 +02:00
Illia Silin
3e6d21adeb enable gfx115x support (#2065) 2025-04-09 10:06:42 -07:00
MHYang-gh
03ce8729fd Make buffer coherence configurable in tensor view (#2041) 
* Make buffer coherence configurable in tensor view

* Fix clang-format for tensor_view.hpp
 2025-04-08 15:34:11 -07:00
Max Podkorytov
6ce0797dad simplify generate_tuple (#2043) 2025-04-08 09:00:51 -07:00
aledudek
80aae6119b [CK_TILE] Fix GEMM Memory Pipeline (#2034) 
* [CK_TILE] Fix GEMM Memory Pipeline

* Fix transpose tile

* Add comments
 2025-04-08 12:40:04 +02:00