Commit Graph

891 Commits

Author SHA1 Message Date
Tianxing Wu
07a0dcd688 Merge branch 'develop' into tianxing/unified-attention 2025-12-02 10:58:31 +00:00
Yi DING
f211156ce6 [CK_Tile] Flatmm MX Cleanup & Explicite Offset Calculation (#3286) 2025-12-02 14:21:12 +08:00
Erwin Terpstra
46f1d740f0 Add grouped gemm instances for RDNA4 (#3237)
* wip: grouped_gemm implementation based on wmma kernel + example for fp16

* chore: clean up grouped_gem_wmma_splitk_fp16 example

* chore: add cmake options to fully disable XDL or WMMA kernels

* feat: add tests for grouped gemma wmma instances for f16 and bf16 (all layouts)

* chore: add grouped gemm wmma bf16 example

* refactor: reuse more code between instance factory functions

* chore: turn test failure if not all batch sizes are supported into a warning

* chore: made failing of test on unsupported instances conditional to not break old tests

* chore: add log message to failure case where AK1/BK1/KBatch is too high for K value

* fix: issue with new overloads of GridwiseGemm_wmma_cshuffle_v3::Run()

* fix: stray comma after parameter list

* fix: compilation issues on RDNA3 and tests failing due to unsupported problems still being ran

* chore: update copyright in header comments

* nit: minor feebdack

* refactor: unified XDL / wma tests

* fix: properly disable FP8 instances when ONLY targeting gfx11

* refactor: add v3 suffix to grouped_gemm device struct name

* fix: small typos in example code

* fix: fully exclude xdl/wmma instances when using the corresponding cmake flags

* chore: remove unused destructor and added pipeline support checks to remove unnecessary paths

* fix: make sure to not add instance library to group if library was skipped

* fix: make sure xdl grouped gemm doesnt fail the new test

* fix: explicitly exclude test if no xdl/wmma support, as pattern matching fails in this case

* fix: examples not working since dependent types and functions were moved to ck namespace in develop

* fix: tests failing when compiling for just gfx11 due to trying to run unsupported instances

* chore: replace/add copyright headers with new format
2025-12-01 15:32:10 -08:00
Cong Ma
23fb253c4e Make CK TILE GEMM Aquant support block tile 128x128x128 (#3325)
* [CK TILE GEMM Quant] Rename GemmConfigBQuantPrefill to GemmConfigQuantPrefill in examples

* [CK TILE GEMM Quant] update tile distribution of aquant

* [CK TILE GEMM Quant] update aquant register offset calculation

* [CK TILE GEMM Quant] Reimplement aquant register offset calculation

* [CK TILE GEMM Quant] Add more unit tests of Aquant

- Test M128xN128xK128

* [CK TILE GEMM Quant] Add more comments to Gemm Aquant
2025-12-01 15:04:37 -08:00
Tianxing Wu
6ecbd7c831 Merge branch 'develop' into tianxing/unified-attention 2025-12-01 11:03:33 +00:00
Aviral Goel
004784ef98 chore(copyright) update library wide CMakeLists.txt copyright header template (#3313)
* chore(copyright) update library wide CMakeLists.txt files copyright header template

* Fix build

---------

Co-authored-by: Sami Remes <samremes@amd.com>
2025-11-28 13:49:54 -08:00
Sami Remes
f981554c39 [CK_TILE] Fix Quant GEMM build (#3320)
* Fix build

* Fix ck_tile example 38 & 40

---------

Co-authored-by: Yi DING <yi.ding@amd.com>
2025-11-28 20:33:53 +08:00
Tianxing Wu
8d25f267ad Merge branch 'tianxing/unified-attention' of https://github.com/ROCm/composable_kernel into tianxing/unified-attention 2025-11-28 12:04:31 +00:00
Cong Ma
30727c48fc Tile engine for streamk (#3157)
* [CK TILE STREAMK] Introduce initial support for tile engine in streamk GEMM.

- This commit lays the groundwork for integrating the tile engine into streamk GEMM.
  It focuses on creating benchmark executables for streamk GEMM.
- Additional scripts like test_benchmark.sh and gemm_benchmark.py will be added once
  the streamk implementation reaches stability.

* [CK TILE STREAMK] Enable CI to execute tile engine benchmarks for StreamK GEMM

* [CK TILE STREAMK] Refactor: Extract common utility functions.

* [CK TILE STREAMK] Revise tile engine of streamk to align with the updated implementation

* Add pre-commit

* [CK TILE STREAMK] Add 'dp_persistent' and 'reduction_strategy' in output of CK TILE STREAMK

* [CK TILE STREAMK] Fix a bug about value of 'dp_persistent' of CK TILE STREAMK

* [CK TILE STREAMK] Update Jenkinsfile

* [CK TILE Engine] Update StreamK tile engine help message

Remove default value messages as they are automatically printed

* [CK TILE Engine] Update StreamK tile engine

- Remove namespace reboot

* [CK TILE Engine] Update StreamK tile engine

- Fix merge error
2025-11-27 15:49:57 -07:00
arai713
24d88d2472 [CK_TILE] Move DataTypeTraits into a Common File (#3146)
This renames the typeToStr struct in the common utilities to DataTypeTraits and removes all duplication of DataTypeTraits across files in CK Tile.

Co-authored-by: Christopher Millette <63608002+cgmillette@users.noreply.github.com>
2025-11-27 09:09:54 -08:00
Tianxing Wu
60ca9484b4 refined benchmarking 2025-11-27 15:07:03 +00:00
Tianxing Wu
eeb419845d fmha v3 flops calculation 2025-11-27 10:32:28 +00:00
Tianxing Wu
c641d0d42c non zero calculation fix 2025-11-27 09:24:52 +00:00
Tianxing Wu
6a2ac8f758 causal mask fix 2025-11-27 09:16:30 +00:00
Matthias Gehre
678298d4c7 Add support for gfx1153 (#3306) 2025-11-27 08:48:00 +01:00
Max Podkorytov
79aae7c7f7 [CK Tile] enable building examples by default (#3259)
* remove EXCLUDE_FROM_ALL from ck-tile examples
-> +15 min build time w/ 64 threads for a single arch

* fix cpp17 compile error in the ck-tile examples

---------

Co-authored-by: khuagarw <khuagarw@amd.com>
Co-authored-by: Ding, Yi <yi.ding@amd.com>
2025-11-26 16:24:44 -08:00
Aviral Goel
35a4b26af0 fix: add dynamic selection of pipelines for aquant mode (#3282)
- Add conditional selection to use v3 pipeline when PreshuffleQuant is true
- Add static assertion in memory pipeline to prevent PreshuffleQuant usage
- Restore BaseBQuantGemmPipelineAgBgCrCompV3 for BQuant cases
- Update BaseGemmPipeline selection to handle all quant modes properly
2025-11-26 10:58:09 +04:00
Aviral Goel
cd47293869 chore(copyright): update copyright header for experimental & example directory (#3292) 2025-11-26 03:09:39 +04:00
Bartłomiej Kocot
00dfa2f2ce [CK TILE] Grouped Conv Explicit Gemm (#3289)
* [CK TILE] Grouped Conv Explicit Gemm

* fixes

* apply builder fixes
2025-11-25 23:28:35 +01:00
Khushbu Agarwal
37ea160088 [CK-Tile] fix block scale example for gfx1201 (#3283) 2025-11-25 13:10:28 -08:00
Bartłomiej Kocot
9ac2666d5b [CK_BUILDER] Add grouped conv bwd ck tile traits (#3281)
* [CK_BUILDER] Add grouped conv bwd ck tile traits

* copilot fixes
2025-11-25 14:57:43 +01:00
Tianxing Wu
cc7caf4d7d correct results 2025-11-25 09:27:40 +00:00
Aviral Goel
d85f065b15 chore(copyright): update copyright header for example directory (#3273)
* chore(copyright): update copyright header for codegen directory

* chore(copyright): update copyright header for example directory
2025-11-24 18:02:41 -08:00
rocking
229d43ea0c Fix batch prefill compile fail in aiter (#3279)
* Fix batch prefill aiter compile fail

* Fix compile error
2025-11-25 09:46:32 +08:00
Thomas Ning
de6a9590ab Reorganize of KPack in GEMM (#3247)
* add the reorganize of KPack

* fix the compilation error

* fix the compilation error
2025-11-24 12:38:59 -08:00
Khushbu Agarwal
8111572785 [CK_Tile] Support for preshuffle weight(B) quant tensor for block scale gemm (#3165)
* formatted

* formatted

* formatting

* formatting

* formatting

* [CK TILE GEMM] Refactor block_scale_gemm examples

- Split cpp file to reduce building time
- Support multiple GemmConfig

* [CK TILE GEMM] Refactor block_scale_gemm examples

- Update Readme

* enable prefill shapes

* [CK TILE GEMM] Refactor block_scale_gemm examples

- Add support for rowcol and tensor GEMM operations

* [CK TILE GEMM] Refactor block_scale_gemm examples

- Update README

* adding preshuffle quant as new parameter and its associated new files

* remove debugging statements

* adding test

* enable preshuffle quant with permuteN

* updating readme and correcponding gemmconfigs

* updating cmake file

* fixing CI failures for grouped quant gemm

* addressing review comments

* fixing CI issue

* addressing reveiw comments

* formatting

* formatting

* fixing aquant operator overlaoding

* formatting

---------

Co-authored-by: Cong Ma <congma13@amd.com>
Co-authored-by: Thomas Ning <Thomas.Ning@amd.com>
2025-11-24 07:48:42 -08:00
Tianxing Wu
b3c5cd0c76 Fixed the block_table 2025-11-24 15:32:33 +00:00
Juuso Korhonen
f2fbc44b7b fix 2025-11-24 10:20:04 +00:00
rocking
5948dbffe4 Support fp8 dynamic quantization for fmha (#3206)
* Support qscale for dynamic quant, remove static quant

* Support hdim=256

* Remove bias test case for fp8

---------

Co-authored-by: Illia Silin <98187287+illsilin@users.noreply.github.com>
Co-authored-by: asleepzzz <hanwen.chang@amd.com>
2025-11-24 16:28:25 +08:00
Johannes Graner
096f0a3b23 [CK Tile] Fix example for conv fwd + bias + clamp (#3235)
* Fix clamp not being applied correctly

* Apply group offsets to D tensors

---------

Co-authored-by: Adam Osewski <19374865+aosewski@users.noreply.github.com>
2025-11-24 07:36:26 +01:00
Emily Martins
2e4b8a8fc4 [CK_TILE] Remove Old CK Tile Stream-K Artifacts (#3202)
* Remove old CK Tile Stream-K implementation

The original CK Stream-K implementation was based on old CK's Stream-K
block to C tile map. However, this implementation did not align with the
original Stream-K paper. Thus, we implemented a new tile partitioner and
associated Stream-K kernel, which was placed in the reboot namespace.

Now that the new Stream-K implementation is ready, this change removes
all artifacts of the old implementation. Specifically, the following
changes were made:
- Removes old Stream-K tile partitioner from CK Tile
- Removes the reboot namespace such that the new implementation resides
  in the ck_tile namespace only.
- Adds tests for bf8 and fp8 using the new implementation
- Removes tests for the old implementation
- Remove the v2 suffix from the new CK Tile Tile Partitioner
derived classes.
- Updates Stream-K Kernel ops file to use /** commenting style.

* Remove v2 from tile partitioner validation function names
2025-11-20 09:32:32 -07:00
asleepzzz
5adaa201ed Revert "Add attn sink (#2892)" (#3250)
This reverts commit 9fa4e8d5ab.
2025-11-20 07:55:15 -08:00
Tianxing Wu
f552cd7841 ref data copying 2025-11-20 11:34:39 +00:00
Linjun-AMD
9fa4e8d5ab Add attn sink (#2892)
* enable attn sink

Signed-off-by: JL-underdog <Jun.Lin@amd.com>

* update attn_sink script

Signed-off-by: JL-underdog <Jun.Lin@amd.com>

* fix some error

Signed-off-by: JL-underdog <Jun.Lin@amd.com>

* clang-format

Signed-off-by: JL-underdog <Jun.Lin@amd.com>

* update fmha_bwd mask

Signed-off-by: JL-underdog <Jun.Lin@amd.com>

* update fmha_bwd_kernel'mask

Signed-off-by: JL-underdog <Jun.Lin@amd.com>

* update block_fmha_pipeline_qr_ks_vs.hpp

Signed-off-by: JL-underdog <Jun.Lin@amd.com>

* fix ci error

Signed-off-by: LJ-underdog <Jun.Lin@amd.com>

* fix format error

Signed-off-by: LJ-underdog <Jun.Lin@amd.com>

* Update block_fmha_bwd_pipeline_default_policy.hpp

* Update fmha_fwd_runner.hpp

* Update block_fmha_batch_prefill_pipeline_qr_ks_vs_async.hpp

* Update fmha_fwd_runner.hpp

* Update fmha_fwd_runner.hpp

* Update fmha_fwd_runner.hpp

* update splitkv_pipline

Signed-off-by: LJ-underdog <Jun.Lin@amd.com>

* update splitkv&pagedkv pipeline

Signed-off-by: LJ-underdog <Jun.Lin@amd.com>

* add sink test

Signed-off-by: LJ-underdog <Jun.Lin@amd.com>

* update attn_sink result log

Signed-off-by: LJ-underdog <Jun.Lin@amd.com>

* update smoke_test_fwd_sink.sh

Signed-off-by: LJ-underdog <Jun.Lin@amd.com>

* update test file

Signed-off-by: LJ-underdog <Jun.Lin@amd.com>

* update test script

Signed-off-by: LJ-underdog <Jun.Lin@amd.com>

* Update block_fmha_fwd_splitkv_pipeline_qr_ks_vs.hpp

* use constexpr kHasSink for sink in fmha pipeline

Signed-off-by: Linjun-AMD <Jun.Lin@amd.com>

* update by pre-commit

Signed-off-by: Linjun-AMD <Jun.Lin@amd.com>

* Update include/ck_tile/ops/fmha/pipeline/block_fmha_pipeline_qr_ks_vs.hpp

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

* Update include/ck_tile/ops/fmha/pipeline/block_fmha_pipeline_qr_ks_vs.hpp

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

* Update include/ck_tile/ops/fmha/kernel/fmha_fwd_pagedkv_kernel.hpp

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

* Update fmha_fwd.py

* Update example/ck_tile/01_fmha/codegen/ops/fmha_fwd_splitkv.py

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

* Update include/ck_tile/ops/fmha/pipeline/block_fmha_fwd_splitkv_pipeline_nwarp_sshuffle_qr_ks_vs.hpp

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

* Remove causal mask setting logic from mask.hpp

Removed the mask setting logic for causal masks.

* fix ci error that some usage of lamada not support in c++17

Signed-off-by: LJ-underdog <Jun.Lin@amd.com>

* Update remod.py

* add smoke sink test

Signed-off-by: LJ-underdog <Jun.Lin@amd.com>

* Update fmha_pagedkv_prefill.py

* Update FmhaFwdPipeline parameters in fmha_fwd.py

* update block_fmha_pipeline_qr_ks_vs_async_trload.hpp

Signed-off-by: LJ-underdog <Jun.Lin@amd.com>

* fix c++17 unsupprot error

Signed-off-by: LJ-underdog <Jun.Lin@amd.com>

* Update block_fmha_fwd_pagedkv_pipeline_qr_ks_vs.hpp

* Fix formatting of sink_seq_end assignment

* Fix indentation for sink_seq_end assignment

* Update block_fmha_fwd_pagedkv_pipeline_qr_ks_vs.hpp

---------

Signed-off-by: JL-underdog <Jun.Lin@amd.com>
Signed-off-by: LJ-underdog <Jun.Lin@amd.com>
Signed-off-by: Linjun-AMD <Jun.Lin@amd.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
2025-11-20 19:24:05 +08:00
Yi DING
47e2ed838e [CK_TILE] Add Flatmm MX FP8 (#3208)
* Use async for flatmm mxfp4

* Fix preshuffle

* Add flatmm mxfp8

* Thanks, Copilot

* Thanks Copilot again~
2025-11-20 10:35:15 +08:00
Yashvardhan Agarwal
1eb26460aa [ck_tile] Pooling example - Improved tile sizes (#3233)
* improved tile sizes

- modified tile sizes for improved example performance

* Update example/ck_tile/36_pooling/pool3d.cpp

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

---------

Co-authored-by: Adam Osewski <19374865+aosewski@users.noreply.github.com>
2025-11-19 15:30:18 +01:00
John Shumway
ad57f6ef0b [CK_BUILDER] Put global CK functions in an the CK namespace (#3232)
* Wrap ck host utitlies in CK namespace.

The CK and CK-Tile source code bases are incompatible because CK is not properly using namespaces everywhere. In particular, we need to put hip_check_error in the ck namespace.

Move all functions in include/ck_/host_utility that were in global namespace into the ck namespace.

There may be additional namespace problems like this, and it's possible we'll have namespace clashes. But it is good design to properly guard our to code bases (CK and CKTile) so that they can both coexist. Moreover, estabilishing this compatiblity is essential if we are going to allow the builder to instantiate  kernels from either template library.

* Add using declarations to test code.

After moving some of the untils into the ck namespace, most examples and a few tests had to be updated to recognize the new namespace declarations. We add using declarations to individual compute units for functions that were previously in the global namespace.

* Add using declarations to client examples.
2025-11-19 11:23:02 +01:00
Aviral Goel
ac70206b2c feat: add support for bf16 for grouped_gemm & grouped_gemm_preshuffle… (#3225)
* feat: add support for bf16 for grouped_gemm & grouped_gemm_preshuffle kernel(s) along with unit test

* docs: Update CHANGELOG.MD
2025-11-18 09:32:27 -05:00
Tianxing Wu
de995fea71 Various fixes 2025-11-18 13:04:58 +00:00
Juuso Korhonen
6ef0b9da8c fixing 2025-11-18 08:57:30 +00:00
Yi DING
b6720531de [CK_TILE] MX Flatmm Split kernel instances (#3207)
* [CK_TILE] MX Flatmm Split kernel instances

* Fix flatmm example compile
2025-11-18 13:46:30 +08:00
Tianxing Wu
8f44fc9593 mem calculation fixed 2025-11-17 14:49:09 +00:00
Tianxing Wu
ff28bd21ba flops and mem calculation 2025-11-17 13:55:54 +00:00
Tianxing Wu
d3c5faf47e Assert block_size num_queries_per_kv 2025-11-17 12:40:31 +00:00
Tianxing Wu
9b68bbd425 Merge branch 'tianxing/unified-attention' of https://github.com/ROCm/composable_kernel into tianxing/unified-attention 2025-11-17 10:06:05 +00:00
Tianxing Wu
5e2fd848b9 remove unneeded args 2025-11-17 10:04:30 +00:00
Juuso Korhonen
5d2a9e5f16 deving the test... 2025-11-17 09:46:31 +00:00
Juuso Korhonen
5e43fd2dfc refactor to clearer BLOCK Q logic 2025-11-17 08:27:19 +00:00
Juuso Korhonen
57a0ec8cc1 add handling for -1 k heads arg 2025-11-17 07:36:42 +00:00
Juuso Korhonen
4a13749f7f fix to example 2025-11-17 07:33:10 +00:00