Rostyslav Geyyer
94c0cadc92
Add instances for grouped conv fwd 3d with ConvScale for fp8@bf8->fp8 ( #1325 )
...
* Add fp8 bf8 conv example
* Add instances
* Add client example
* Add random scale values
* Format
[ROCm/composable_kernel commit: acda4c5a3c
2024-06-12 14:41:56 -05:00
Rostyslav Geyyer
25ae51c6f0
Add a convinvscale op, related instances and examples ( #1307 )
...
* Update the element op
* Add an example
* Add instances
* Add a client example
* make sure new instances only build on gfx9
* Update element op and its handling
* Format
* Update instances to take element op as an argument
* Update examples to use random scale values
* Format
* Update client example with random scales
* Format
---------
Co-authored-by: illsilin <Illia.Silin@amd.com >
[ROCm/composable_kernel commit: ce66277a76
2024-06-10 14:48:49 -05:00
Bartłomiej Kocot
41c68496e6
Integrate universal gemm with conv forward ( #1320 )
...
* Integrate universal gemm with conv fwd
* Fix conv fwd wmma test
* Fix instances
* Remove direct load check
[ROCm/composable_kernel commit: ac58cc5d1d
2024-06-05 13:01:29 -05:00
Rostyslav Geyyer
fec15d8c40
Add a scale op, related instances and examples ( #1242 )
...
* Add a scale op
* Update the element op
* Add instances
* Add an example
* Add a client example
* Add a flag check
* Revert flag check addition
* Fix flag check
* Update d strides in example
* Update d strides in client example
* Apply suggestions from code review
Update copyright header
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com >
* Move the example
* Move the client example
* Update element op
* Update example with the new element op
* Add scalar layout
* Update example
* Update kernel for scalar Ds
* Revert kernel changes
* Update element op
* Update example to use scales' pointers
* Format
* Update instances
* Update client example
* Move element op to unary elements
* Update element op to work with values instead of pointers
* Update instances to take element op as an argument
* Update examples to use random scale values
---------
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com >
[ROCm/composable_kernel commit: cb0645bedc
2024-06-04 19:28:15 -05:00
Illia Silin
9bdae6116c
Split the gemm_multi_abd instances. ( #1306 )
...
* split the gemm_multi_abd instances
* update the dates
[ROCm/composable_kernel commit: ec2bae27ff
2024-05-23 09:17:02 -07:00
Bartłomiej Kocot
b4b436d29a
Optimize grouped conv bwd weight for small M and N ( #1303 )
...
* Optimize grouped conv bwd weight for small M and N
* Fixes
[ROCm/composable_kernel commit: fd72380aeb
2024-05-22 21:01:01 +02:00
Illia Silin
beb7927f52
Select appropriate GPU targets for instances, tests, and examples. ( #1304 )
...
* set individual gpu targets for instances, examples, tests
* fix path to hip compiler
* fix path to hip compiler once more
* aggregate device macros in ck_tile config header
* fix the cmake logic for instances
* fix clang format
* add gfx900 and gfx906 to default set of targets
[ROCm/composable_kernel commit: 7b027d5643
2024-05-22 11:45:27 -07:00
Bartłomiej Kocot
b6a17bc3e2
Add two stage grouped conv bwd weight kernel ( #1280 )
...
[ROCm/composable_kernel commit: 0b6b5d1785
2024-05-08 09:53:24 +02:00
Rostyslav Geyyer
38d64386cd
Add an ignore ( #1270 )
...
[ROCm/composable_kernel commit: a2d0bdd5a9
2024-04-30 20:45:22 -07:00
Rostyslav Geyyer
e0669ecf6c
Mark unneeded instances as "getting deprecated" ( #1265 )
...
* Add a flag
* Add flag check and messages
---------
Co-authored-by: root <root@aus-g7-rogeyyer.amd.com >
[ROCm/composable_kernel commit: 6ced3c12ff
2024-04-29 12:00:55 -07:00
Haocong WANG
1a34c500a6
[GEMM] UniversalGemm update ( #1262 )
...
* Add bf16 instances
* Add bf16 gemm universal example
* tempsave
* Add guard to navi compilation
* workground on a specific mixed gemm instance ( bring back it when compiler fix upload)
* fix formatting condition statement issue
* solve conflict
---------
Co-authored-by: Jun Liu <Liu.Jun@amd.com >
[ROCm/composable_kernel commit: 764164b488
2024-04-26 12:56:07 -05:00
zjing14
58142fae83
ggemm tile_loop multD bf16 int8 ( #1258 )
...
* Overload output stream operator for LoopScheduler and PiplineVersion
* Add Run overload accepting grid descriptors MK.
* Add __device__ keyword for CalculateGridSize
* Create device op GroupedGemmMultipleD
* Add GroupedGemm MultipleD Tile Loop implementation.
* Add an example for GroupedGemm MultipleD tile loop.
* Device Op GroupedGEMMTileLoop.
* Bunch of small changes in exmaple.
* CkProfiler
* Remove unused tparam.
* changed the copy function to v7r2
* adding multi_abd
* in-progress
* add post-load oob check
* Fix include statement.
* Fix output stream overloads.
* Do not make descriptors and check validity untill we find group.
* Fix gemm desc initialization.
* debugging
* adjust instances
* add run_lds
* add elemntwise_op
* replace multi_abd_device with v3
* clean up
* clean
* clean
* Revert device op
* Fix compilation for DTYPES=FP16
* Validate tensor transfers paramters.
* Added LDSType
* profiling
* adjust oobcheck
* add missing file
* Validate on host only NK dims if M is not known.
* add
* clean
* refactor
* clean
* add examples
* add fuse
* add fusion and client example
* Fix bug.
* A convenient debug func for selecting threads.
* Fix has main k block loop bug.
* Make sure that b2c has up to date tile offset.
* Output stream operator for Sequence type.
* Cmake file formatting.
* clean
---------
Co-authored-by: Adam Osewski <Adam.Osewski@amd.com >
[ROCm/composable_kernel commit: 5ae893c0d3
2024-04-26 10:37:49 -05:00
zjing14
ce67c185b4
bf16A_Int8B with fastgelu/bias ( #1264 )
...
* changed the copy function to v7r2
* adding multi_abd
* in-progress
* add post-load oob check
* debugging
* adjust instances
* add run_lds
* add elemntwise_op
* replace multi_abd_device with v3
* clean up
* clean
* clean
* Added LDSType
* profiling
* adjust oobcheck
* add missing file
* refactor
* clean
* add examples
[ROCm/composable_kernel commit: 0d0150db20
2024-04-26 07:26:30 -05:00
Adam Osewski
2b452ad135
Grouped GEMM Multiple D tile loop. ( #1247 )
...
* Overload output stream operator for LoopScheduler and PiplineVersion
* Add Run overload accepting grid descriptors MK.
* Add __device__ keyword for CalculateGridSize
* Create device op GroupedGemmMultipleD
* Add GroupedGemm MultipleD Tile Loop implementation.
* Add an example for GroupedGemm MultipleD tile loop.
* Device Op GroupedGEMMTileLoop.
* Bunch of small changes in exmaple.
* CkProfiler
* Remove unused tparam.
* Fix include statement.
* Fix output stream overloads.
* Do not make descriptors and check validity untill we find group.
* Fix gemm desc initialization.
* Revert device op
* Fix compilation for DTYPES=FP16
* Validate tensor transfers paramters.
* Validate on host only NK dims if M is not known.
* Fix bug.
* A convenient debug func for selecting threads.
* Fix has main k block loop bug.
* Make sure that b2c has up to date tile offset.
* Output stream operator for Sequence type.
* Cmake file formatting.
[ROCm/composable_kernel commit: b4032629e5
2024-04-25 15:12:53 -05:00
Bartłomiej Kocot
6578635cb3
Refactor elementwise kernels ( #1222 )
...
* Refactor elementwise kernels
* Instances fixes
* Fix cmake
* Fix max pool bwd test
* Update two stage gemm split k
* Restore elementwise scale for hiptensor backward compatiblity
* Fix Acc data type check in conv fwd multiple abd
* Disable conv fp64 fwd example
* Update grouped conv weight multi d
[ROCm/composable_kernel commit: ad1597c499
2024-04-19 13:31:17 +02:00
jakpiase
e7d121a6f0
Add bf16 and bf16@int8 mk_nk_mn instances for grouped gemm two stage ( #1228 )
...
* added bf16 and bf16@int8 mk_nk_mn instances
* fix preprocessor guards
[ROCm/composable_kernel commit: e0f3f918f1
2024-04-19 13:16:10 +02:00
Bartłomiej Kocot
d001bea12f
Add grouped conv bwd weight multi d kernel ( #1237 )
...
* Add grouped conv bwd weight multi d kernel
* Reference fix
* Fix cmake files
* bwd weight scale only xdl
* Fixes
* Fix client conv fwd example
[ROCm/composable_kernel commit: fd923b6d86
2024-04-18 23:35:04 +02:00
zjing14
4ddb546fe5
Added Multi_ABD support into Gemm and GroupedGemmFixedNK ( #978 )
...
* added an example grouped_gemm_multi_abd
* fixed ci
* add setElementwiseOp
* changed API
* clean code: add multiA into example
* fixed v7r2 copy
* add transpose
* clean
* fixed vector_load check
* Update example/15_grouped_gemm/grouped_gemm_multi_abd_xdl_fixed_nk_bias_fp16.cpp
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com >
* Update example/15_grouped_gemm/grouped_gemm_multi_abd_xdl_fixed_nk_bias_fp16.cpp
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com >
* Update example/15_grouped_gemm/grouped_gemm_multi_abd_xdl_fixed_nk_bias_fp16.cpp
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com >
* Update include/ck/tensor_operation/gpu/grid/gridwise_gemm_multiple_abd_xdl_cshuffle.hpp
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com >
* Update include/ck/tensor_operation/gpu/grid/gridwise_gemm_multiple_abd_xdl_cshuffle.hpp
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com >
* Update include/ck/tensor_operation/gpu/device/device_grouped_gemm_multi_abd.hpp
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com >
* Update include/ck/tensor_operation/gpu/device/device_grouped_gemm_multi_abd.hpp
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com >
* Update include/ck/tensor_operation/gpu/device/device_grouped_gemm_multi_abd.hpp
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com >
* Update include/ck/tensor_operation/gpu/device/device_grouped_gemm_multi_abd.hpp
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com >
* Update include/ck/tensor_operation/gpu/device/device_grouped_gemm_multi_abd.hpp
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com >
* Update include/ck/tensor_operation/gpu/device/device_grouped_gemm_multi_abd.hpp
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com >
* Update include/ck/tensor_operation/gpu/device/device_grouped_gemm_multi_abd_fixed_nk.hpp
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com >
* Update include/ck/tensor_operation/gpu/device/device_grouped_gemm_multi_abd_fixed_nk.hpp
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com >
* add reduce
* testing
* add example_b16_i8
* refactor example
* clean
* add mpading
* disable reduce for kbatch = 1
* seperate reduce device op
* add reduce op
* add guard for workspace_size
* add instances
* format
* fixed
* add client example
* add a colmajor
* add instances
* Update cmake-ck-dev.sh
* Update profile_gemm_splitk.cpp
* Update gridwise_gemm_xdlops_v2r4r2.hpp
* format
* Update profile_gemm_splitk.cpp
* fixed
* fixed
* adjust test
* adjust precision loss
* adjust test
* fixed
* add bf16_i8 scale bias
* fixed scale
* fixed scale elementwise_op
* revert contraction deviceop changes
* fixed
* Add AddFastGelu
* Revert "Merge branch 'jizhan/gemm_splitk_reduce' into grouped_gemm_multi_abd_fixed_nk_example"
This reverts commit 3b5d001efd943199a991barkocot@amd.com >
* Update device_reduce_threadwise.hpp
* clean
* Fixes
* Fix example
---------
Co-authored-by: Jing Zhang <jizha@amd.com >
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com >
[ROCm/composable_kernel commit: 12865fbf28
2024-04-15 21:09:45 -05:00
Haocong WANG
ceaecc86ca
[GEMM] Gemm universal device operation ( #1154 )
...
* Optimize GEMM on MI200/300:
1. Add new blockwise gemm pipeline
2. Add irregular splitk intances
* clang format + typo fix
* Fix a bug
* initial commit
* Add more instances to irregular splitk
* blkgemm pipeline v1~4 prototype
* Sanity Checked. Known issue:
1. Poor performance of splitk
2. Register spill on blkgemmpipeline v3
* Sanity and Performance fix:
1. fix a bug related to sanity in grouped b2c mapping
2. fix a bug related to sanity and performance in splitk offset
* Sanity and API update:
1. Remove prefetch stage
2. Fix valid check bug
3, Add first gemm_universal instance into ckProfiler
* Add NN instances for gemm universal
* 1. Add NT instances for gemm_universal
2. Fix a bug about Kpadding in gemm_universal
* Fix a bug regarding padding Odd K number
* remove kernel print
* Fix KPadding bug...
* Update safety check
* another try to fix kpadding..
* Sanity checked
* new instances..
* clang format+typo fix
* remove clang format script's change
* Add non-hotloop compile option
* 1. Add fp16xfp8 example
2. pull packed convert f8 from pr1150
* Some miscs.. opt and fix
* Add pipeline description docs
* Split universal gemm instance library to cut profiler compiling time
* uncomment cmakefile
* Fix a bug caused by blockwise_gemm_pipe_v2
* reduce default splitk to 1
* Add 224x256x64 tile size
* update, including:
1. Experiment pipeline 5~7
2. Optimization for pipeline 4
3. Organized instance library
* temp save
* temp save
* Permuted lds layout, sanity and function checked
* clang format
* Move OOB check from RunRead to RunWrite, for better software pipeline.
TODO: agpr spill when NN layout
* clangformat
* A/B splitpipe scheduler for v3
* Fix two bugs
* bug fix
* fix a bug in oob check
* Example for mixed fp16_fp8 gemm
* Clean experimental code blocks
* Add mixed precision gemm into profiler
* tempsave
* optimize m/n major lds layout
* Add RRR GEMM mixed precision instances
* Optimize f8 matrix transpose
* Add test_gemm_universal
* A/B spilt schedule for blkpip v5
* Take ds_read2 into iglp scheduling scheme
* format
* fixed cmake
* Add llvm-option into CI cmake flag
---------
Co-authored-by: Jing Zhang <jizhan@amd.com >
[ROCm/composable_kernel commit: f83e9701e9
2024-04-13 21:03:18 -05:00
Rostyslav Geyyer
36e4675cc5
Add instances for conv_scale with bf8@fp8->fp8 ( #1231 )
...
* Add instances
* Add example
* Add profiler mode
* Add client example
[ROCm/composable_kernel commit: bbefc12a26
2024-04-11 10:35:00 -05:00
Bartłomiej Kocot
f9db066f3f
Extend support for contraction 6D ( #1207 )
...
* Extend support for contraction up to 5D
* Extend contraction bilinear instances
* Fix interface test
* Add 6d support, remove 3d,4d,5d
* Fixes
* Fix readme
* Make defualt dim for contraction instances
[ROCm/composable_kernel commit: ced5af16f7
2024-04-09 23:46:21 +02:00
jakpiase
1438bdd38c
Add Grouped Gemm Multiple D SplitK TwoStage ( #1212 )
...
* Support A/B/C elementwise ops.
* First part of GGEMM multiD splitk two stage.
* WIP - changes for debuggin.
* tmp save
* working version
* added bf16@int8 version
* fixes
* add reviewers sugestions
* pre-commited missing files
* switched to ifs from elseifs
---------
Co-authored-by: Adam Osewski <Adam.Osewski@amd.com >
[ROCm/composable_kernel commit: c701071666
2024-04-04 11:01:33 +02:00
Rostyslav Geyyer
0b8e766e55
Add instances for conv_scale with fp8@bf8->fp8 ( #1220 )
...
* Update device op api to support BComputeType
* Add example
* Add instances
* Add profiler mode
* Add client example
* Update copyright year
* Add BComputeType check
* Fix compute types
[ROCm/composable_kernel commit: a61e73bc56
2024-04-03 09:08:08 -05:00
Illia Silin
1f4d13b2b5
Split the instances by architecture. ( #1223 )
...
* parse examples inside the add_example_executable function
* fix the example 64 cmake file
* add xdl flag to the gemm_bias_softmax_gemm_permute example
* add filtering of tests based on architecture type
* enable test_grouped_gemm for gfx9 only
* enable test_transpose only for gfx9
* only linnk test_transpose if it gets built
* split the gemm instances by architectures
* split gemm_bilinear,grouped_conv_bwd_weight instances by targets
* split instances by architecture
* split grouped_conv instances by architecture
* fix clang format
* fix the if-else logic in group_conv headers
* small fix for grouped convolution instances
* fix the grouped conv bwd weight dl instances
* fix client examples
* only enable client examples 3 and 4 on gfx9
* set the gfx9 macro
* make sure the architecture macros are set by cmake
* use separate set of xdl/wmma flags for host code
* sinmplify the main cmake file
* add conv_fwd_bf8 instance declaration
[ROCm/composable_kernel commit: ae57e5938e
2024-04-02 09:42:17 -07:00
Bartłomiej Kocot
c3891d6cb1
Add elementwise with dynamic vector dim ( #1198 )
...
* Add elementwise with dynamic vector dim
* Reduce number of instaces
* Fixes
* Fixes
[ROCm/composable_kernel commit: 9c052804a7
2024-03-22 10:40:43 +01:00
Rostyslav Geyyer
4fe9769987
Add instances for conv_scale with bf8 in / fp8 out ( #1200 )
...
* Add bf8 conv fwd instances
* Add example
* Add profiler mode
* Add client example
* Fix copyright headers
* Format
[ROCm/composable_kernel commit: fd0d093e78
2024-03-21 13:57:34 -05:00
Rostyslav Geyyer
17be33ccf9
Add instances for conv_scale with fp8 in/out ( #1193 )
...
* Add fp8 conv instances and client example
* Format
* Add example
* Update cmakelists
* Add profiler mode
* Format
* Fix copyright headers
[ROCm/composable_kernel commit: e626d5202a
2024-03-15 09:50:03 -07:00
Bartłomiej Kocot
6cbbbe7524
Add conv fwd/bwd data scale instances, extend bilinear instances ( #1178 )
...
* Add conv fwd/bwd data scale instances
* Fix cmake client example file
---------
Co-authored-by: Adam Osewski <19374865+aosewski@users.noreply.github.com >
[ROCm/composable_kernel commit: 285251768e
2024-03-13 23:09:08 +01:00
zjing14
304b7c3abb
Navi3 rel ( #1176 )
...
* wmma_op + unit test
* add arch limitation to wmma test
* change arch limitation
* Refactor + Add all type unit test(int4 compile failed)
* Add f32_16x16x16_bf16 unit test
* tempsave
* tempsave
* tempsave
* runtime bug, cannot find symbol
* workaround for incorrect HIP warpSize return value
* debugging
* tempsave
* Correctness OK, waiting for optimization
* Tidy up + format
* temp save
* temp save, reproduce the v_bfi_b32 issue
* add inline asm for wmmaop test
* tidy up
* clean some debug purpose code
* discard some codes
* clang format
* clang format
* compiler issue fixed + increase tile size
* navi3x_multipleD+example
* temp save
* workable
* batchedgemm[OK], groupconv[debug]
* groupconv: Sanity check[OK], Performance[Bad]
* navi3x_groupconv_need_optimization
* create necessary files
* save progress
* Add Inter-Row thread transfer
* save progress
* save debugging progress
* sanity check pass
* fix a host tensor bug and clean up flash-attn code
* format
* cancel unnecessary change
* cancel unnecessary change
* cancel unnecessary change
* temp save, add asm backend flag to amd_wmma
* Mat-A LDS Bypass sanity pass
* temp save
* gemm sanity fix
* Porting new blockwise gemm to flash attention
* Example branch provide to compiler team
* tempsave
* Fix a bug
* batched gemm ported
* conv A-skip lds ported
* Skip B-Lds real gemm
* Skip B Lds Gemm + MulD
* batched gemm, conv, skip b lds
* format
* Attn, skip b lds
* Change GridwiseOp nam
* fix a typo caused bug
* Skip A_Lds sanity pass, Skip B_Lds scratch occured
* Bug found, intra-row permute off caused
* bug found
* a fix
* disable buffer load due to incorrect 3rd dword
* update fmha config, no scratch generated
* update 3rd dword
* fmha config update
* FMHA, add support to gfx1101/gfx1102
* Merge origin dev (#2 )
* [Navi3x] Fix Gridwise_multiple_d operation (#649 )
* Add CMake Option "USE_OPT_NAVI3X"
* fix bug
* standardize docs (#655 )
* Separate bibtex requirement from rocm-docs-core (#656 )
* separate bibtex requirement from rocm-docs-core
* point requirements to source rocm-docs-core repo
* Add CMake Option "USE_OPT_NAVI3X" (#647 )
* Add CMake Option "USE_OPT_NAVI3X"
* remove navi3x opt compile option from cmake script
* Conv + quantization + tanh (#645 )
* Rename file. Prepare to support another activation
* Add comment for quantization
* Extract out_elementop
* Add tanh example
* Add conv + bias + tanh quantization instance
* Add missing parameter
* Refine cmake
* Add external api and client example
* Extract variable in example
* Fix the comment
---------
Co-authored-by: zjing14 <zhangjing14@gmail.com >
* Add a denorm test fix (#603 )
* Add type_convert implementations for bf16
* Add the fix for conv_fwd
* Add the fix for conv_bwd_data
* Add the fix for conv_bwd_weight
* Format
* Format
* Another format
* Add a macro to use workaround on MI200 only
* Format
---------
Co-authored-by: Rosty Geyyer <rosty.geyyer@amd.com >
Co-authored-by: zjing14 <zhangjing14@gmail.com >
* simplify karg in device/grid of split-k op (#644 )
* simplify karg in device/grid split-k op
* fix mk_kn_mn instances
* add more instances
* use name from tensor layout
* fix 3rd dword of buffer source descriptor (#659 )
* add fp64 instances (#658 )
Co-authored-by: root <root@ctr-ubbsmc15.amd.com >
* Issue #666 : Revert "simplify karg in device/grid of split-k op (#644 )" (#665 )
This reverts commit 1108f64591#668 )
* Rename to proper naming
* Add example of groupnorm + swish
* Extract duplicate code in example
* Add groupnorm + swish instances
* Ractor instance generation, split into multiple cpp file
* Add external api and client example
* Refine profiler message
* Use ck math version of exp
* Refine problem size in example
* Add host version of exp
* add a marco to turn on/off denorm fix (off by default) (#673 )
* add a marco to turn off denorm fix by default
* expose the marco
---------
Co-authored-by: root <root@ctr-ubbsmc15.amd.com >
* fixed quant example (#672 )
Co-authored-by: root <root@ctr-ubbsmc15.amd.com >
* Add dependabot config and pin rocm-docs-core (#663 )
* [gtest] suppress unsafe buffer warn (#670 )
ref: https://github.com/ROCmSoftwarePlatform/MIOpen/pull/1912
* Add memory index guard in wmma device ops (#667 )
* Add more macros to turn on/off denorm fix (#678 )
Co-authored-by: Rosty Geyyer <rosty.geyyer@amd.com >
* Fix a typo (#676 )
* Add (#677 )
* Allow using ROCm release candidate compilers. (#679 )
* enable use of rocm5.5 release candidate 4
* upgrade to ROCM5.5 RC5
* try fix the PUB_KEY error, remove the cmake-data package
* upgrade to latest cmake version
* use private dockerhub repo for rocm5.5 rc5
* add missing bracket
* add vector load check
* solve conflicts
---------
Co-authored-by: Sam Wu <sjwu@ualberta.ca >
Co-authored-by: Sam Wu <sam.wu2@amd.com >
Co-authored-by: rocking5566 <ChunYu.Lai@amd.com >
Co-authored-by: zjing14 <zhangjing14@gmail.com >
Co-authored-by: Rostyslav Geyyer <46627076+geyyer@users.noreply.github.com >
Co-authored-by: Rosty Geyyer <rosty.geyyer@amd.com >
Co-authored-by: carlushuang <carlus.huang@amd.com >
Co-authored-by: root <root@ctr-ubbsmc15.amd.com >
Co-authored-by: Jun Liu <Liu.Jun@amd.com >
Co-authored-by: Illia Silin <98187287+illsilin@users.noreply.github.com >
* Disable SkipLDS & Align AIT api (#3 )
* fix layernorm, reduction Ops (#4 )
* [Navi3x] Fix Gridwise_multiple_d operation (#649 )
* Add CMake Option "USE_OPT_NAVI3X"
* fix bug
* standardize docs (#655 )
* Separate bibtex requirement from rocm-docs-core (#656 )
* separate bibtex requirement from rocm-docs-core
* point requirements to source rocm-docs-core repo
* Add CMake Option "USE_OPT_NAVI3X" (#647 )
* Add CMake Option "USE_OPT_NAVI3X"
* remove navi3x opt compile option from cmake script
* Conv + quantization + tanh (#645 )
* Rename file. Prepare to support another activation
* Add comment for quantization
* Extract out_elementop
* Add tanh example
* Add conv + bias + tanh quantization instance
* Add missing parameter
* Refine cmake
* Add external api and client example
* Extract variable in example
* Fix the comment
---------
Co-authored-by: zjing14 <zhangjing14@gmail.com >
* Add a denorm test fix (#603 )
* Add type_convert implementations for bf16
* Add the fix for conv_fwd
* Add the fix for conv_bwd_data
* Add the fix for conv_bwd_weight
* Format
* Format
* Another format
* Add a macro to use workaround on MI200 only
* Format
---------
Co-authored-by: Rosty Geyyer <rosty.geyyer@amd.com >
Co-authored-by: zjing14 <zhangjing14@gmail.com >
* simplify karg in device/grid of split-k op (#644 )
* simplify karg in device/grid split-k op
* fix mk_kn_mn instances
* add more instances
* use name from tensor layout
* fix 3rd dword of buffer source descriptor (#659 )
* add fp64 instances (#658 )
Co-authored-by: root <root@ctr-ubbsmc15.amd.com >
* Issue #666 : Revert "simplify karg in device/grid of split-k op (#644 )" (#665 )
This reverts commit 1108f64591#668 )
* Rename to proper naming
* Add example of groupnorm + swish
* Extract duplicate code in example
* Add groupnorm + swish instances
* Ractor instance generation, split into multiple cpp file
* Add external api and client example
* Refine profiler message
* Use ck math version of exp
* Refine problem size in example
* Add host version of exp
* add a marco to turn on/off denorm fix (off by default) (#673 )
* add a marco to turn off denorm fix by default
* expose the marco
---------
Co-authored-by: root <root@ctr-ubbsmc15.amd.com >
* fixed quant example (#672 )
Co-authored-by: root <root@ctr-ubbsmc15.amd.com >
* Add dependabot config and pin rocm-docs-core (#663 )
* [gtest] suppress unsafe buffer warn (#670 )
ref: https://github.com/ROCmSoftwarePlatform/MIOpen/pull/1912
* Add memory index guard in wmma device ops (#667 )
* Add more macros to turn on/off denorm fix (#678 )
Co-authored-by: Rosty Geyyer <rosty.geyyer@amd.com >
* Fix a typo (#676 )
* Add (#677 )
* Allow using ROCm release candidate compilers. (#679 )
* enable use of rocm5.5 release candidate 4
* upgrade to ROCM5.5 RC5
* try fix the PUB_KEY error, remove the cmake-data package
* upgrade to latest cmake version
* use private dockerhub repo for rocm5.5 rc5
* add missing bracket
* Disable SkipLDS & Align AIT api
* Update dependabot config (#682 )
Co-authored-by: samjwu <samjwu@users.noreply.github.com >
* update attn api
* solve type_convert bug + enable
---------
Co-authored-by: Sam Wu <sjwu@ualberta.ca >
Co-authored-by: Sam Wu <sam.wu2@amd.com >
Co-authored-by: rocking5566 <ChunYu.Lai@amd.com >
Co-authored-by: zjing14 <zhangjing14@gmail.com >
Co-authored-by: Rostyslav Geyyer <46627076+geyyer@users.noreply.github.com >
Co-authored-by: Rosty Geyyer <rosty.geyyer@amd.com >
Co-authored-by: carlushuang <carlus.huang@amd.com >
Co-authored-by: root <root@ctr-ubbsmc15.amd.com >
Co-authored-by: Jun Liu <Liu.Jun@amd.com >
Co-authored-by: Illia Silin <98187287+illsilin@users.noreply.github.com >
Co-authored-by: samjwu <samjwu@users.noreply.github.com >
Co-authored-by: haocwang <Haocong.WANG@amd.com >
* fix typo
* Fix attention with causal mask
* multiple fix, try ait compile
* Add A/B not use LDS pipeline
* Clang format, Add gfx1101, gfx1102 support of FMHA example
* cancel change of format script
* 1. Enable 2-stage global Prefetch ( May cause VGPR spilling)
2. Enable FP16 accumulator blockwise_gemm
* clang-format
* 1. change blockwise gemm loopover direction from kmn to mnk ( ~1% improvement)
2. change kernel timing mode to 50 warmup + 50 timed repeat
* Update low level abstration of blockwise gemm wmma
* (2/5) bilinear gemm pass, perf bug: skip a lds has lower performance than skip b lds
* (3/5) batched gemm pass, perf bug: skip a lds has lower performance than skip b lds
* (4/5) grouped conv pass
* (5/5) attention pass, todo: debug lds perf bug
* AIT Attention API refactor (#8 )
* sanity pass
* sanity pass 2
* confirm significant performance regression.
* turn on all instances
* turn off instance format
* Fix bug & tunning & format
* DML meta, self_attn+cross_attn
* sanity pass
* remove useless flag
* update tile and problem size used in AIT attention
* bug fix in grouped conv supporting check
* deprecate inline asm wmma
* Bug fix: double lds skip
* clang-format
* Fix errors in
1. example, fmha
2. gridwise pipeline
3. deviceop, fmha, change some containers from vector to array
* part2 of previous commit
* clang format
* API fix of gridwisegemmpipeline
* separate array base and vector base attention tensor transformation
* fix gemm
* clang format
* add gemm fp16 instances
* Temp save
* fpAintB kernel compile pass
* Sanity pass.
* Temp save
* debug code enabled
* Fp16AInt8B_GEMM sanity
* MQA implementation
* GQA-4 example
* tempsave
* Compile pass
* New implementation of fp16Aint8B Gemm, Acheieve similar math throughput with native fp16 Gemm
* format
* Todo: fix gemm_bilinear_wmma instances compilation bug
* Solve a bug when K1=16
* remove unnecessary changes
* Remove tensor layout limitation to LDS usage in tesnor contraction
* update self-attention and cross-attention
* fix a typo of name
* Add arch limiter for fp8 gemm
* enable fp8 gemm_xdl for all gfx9 targets
* temporarily disable gemm_xdl_fp16_fp8 on MI100/200
* fix the cmake logic for gemm_xdl_fp16_fp8
* re-enable the gemm_xdl_fp16_fp8 on MI100/200
---------
Co-authored-by: aska-0096 <haocwang@amd.com >
Co-authored-by: Sam Wu <sjwu@ualberta.ca >
Co-authored-by: Sam Wu <sam.wu2@amd.com >
Co-authored-by: rocking5566 <ChunYu.Lai@amd.com >
Co-authored-by: Rostyslav Geyyer <46627076+geyyer@users.noreply.github.com >
Co-authored-by: Rosty Geyyer <rosty.geyyer@amd.com >
Co-authored-by: carlushuang <carlus.huang@amd.com >
Co-authored-by: root <root@ctr-ubbsmc15.amd.com >
Co-authored-by: Jun Liu <Liu.Jun@amd.com >
Co-authored-by: Illia Silin <98187287+illsilin@users.noreply.github.com >
Co-authored-by: samjwu <samjwu@users.noreply.github.com >
Co-authored-by: haocwang <Haocong.WANG@amd.com >
Co-authored-by: illsilin <Illia.Silin@amd.com >
[ROCm/composable_kernel commit: 1837040a9c
2024-03-08 17:11:51 -08:00
jakpiase
b3a942c03c
Add support for mixed precision bf16&int8 grouped gemm ( #1166 )
...
* add support for mixed precision bf16&int8 grouped gemm
* fix gfx versions and add bf16 kbatch condition
* added reviewers comments
[ROCm/composable_kernel commit: 32d4be3d09
2024-02-21 10:35:35 +01:00
Bartłomiej Kocot
4d035536cb
Extend permute scale support up to 6D ( #1168 )
...
* Extend permute scale support up to 6D
* Fixes
* Fixes
* Update profiler/README.md
Co-authored-by: Lisa <lisajdelaney@gmail.com >
* Update profiler/README.md
Co-authored-by: Lisa <lisajdelaney@gmail.com >
* Update profiler/README.md
Co-authored-by: Lisa <lisajdelaney@gmail.com >
* Update profiler/README.md
Co-authored-by: Lisa <lisajdelaney@gmail.com >
* Update profiler/README.md
Co-authored-by: Lisa <lisajdelaney@gmail.com >
* Update profiler/README.md
Co-authored-by: Lisa <lisajdelaney@gmail.com >
* Update profiler/README.md
Co-authored-by: Lisa <lisajdelaney@gmail.com >
---------
Co-authored-by: Lisa <lisajdelaney@gmail.com >
[ROCm/composable_kernel commit: 66736edb95
2024-02-20 09:56:54 -08:00
Bartłomiej Kocot
34bb3c7e31
Add bilinear conv fwd and bwd data instances ( #1164 )
...
[ROCm/composable_kernel commit: bf98b47697
2024-02-13 11:49:05 +01:00
zjing14
8063a407e7
Optimizing fp8_fp16 mixedprec gemm ( #1150 )
...
* add delayed cvt
* extend fp16 gemm_splitk instances for fp8_fp16 gemm
* add f8 example
* add 128 kperblk instances for fp8
* add kpb128 instance
* added more instances into kpb128
* clean code
* clean code
* fix
* fix
* fixed
* Update example/35_splitK_gemm/splitK_gemm_xdl_fp16_fp8.cpp
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com >
* Update include/ck/tensor_operation/gpu/thread/threadwise_tensor_slice_transfer.hpp
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com >
* Update library/src/tensor_operation_instance/gpu/gemm_splitk/device_gemm_xdl_splitk_f16_fp8_f16_mk_nk_mn_kpb128_instance.cpp
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com >
---------
Co-authored-by: Jing Zhang <jizha@amd.com >
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com >
[ROCm/composable_kernel commit: 602c4cc0d9
2024-02-12 09:45:42 -08:00
zjing14
cc25dc316f
add generic instances for DeviceGemm_Xdl_CShuffle ( #1161 )
...
* add generic instances
* clean code
[ROCm/composable_kernel commit: 94fbaac002
2024-02-09 10:20:53 -06:00
Illia Silin
0a90ec2237
Split-up instances to improve build times. ( #1159 )
...
* split up splitk-gemm instances
* clean up some unused variables
* split the mk_kn_mn interwave splitk-gemm instances
* split up f16_f16_f16 mk_nk_mn splitk gemm instances
* fix clang format
* fix function names
* fix typo
* split up the 2 largest fp16*fp8 splitk gemm instances
* get rid of unused variables
* split up the largest splitk-gemm fp8*fp16 instance file
* split up the instances for xdl fp8 gemms
* split the headers for f16 and i8 for wmmma convolution instances
[ROCm/composable_kernel commit: 1b0fbaebbb
2024-02-07 12:47:12 -08:00
jakpiase
b41cf51951
Add support for mixed-precision f16bf16_int8 gemm ( #1127 )
...
[ROCm/composable_kernel commit: ba86eadce5
2024-02-07 15:54:13 +01:00
Bartlomiej Wroblewski
dd3c37d031
Implement direct loads split-K GEMM kernel ( #1137 )
...
* WIP: Implement direct loads split-K GEMM kernel
* Clean the review
---------
Co-authored-by: Adam Osewski <19374865+aosewski@users.noreply.github.com >
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com >
[ROCm/composable_kernel commit: 6951858221
2024-02-07 01:08:34 +01:00
rocking
133d8e762f
layernorm & groupnorm bwd gamma beta ( #1133 )
...
* Add layernorm bwd gamma beta external api
* Add groupnorm external api
* Add layernorm bwd gamma beta profiler
* Add groupnorm bwd gamma beta ckProfiler
* Add layernorm & groupnorm bwd gamma beta test
* Fix groupnorm bwd gamma beta profiler bug
* Layernorm bwd weight client example
* Groupnorm bwd weight client example
* clang format
* Remove useless header
* Let inv_std be positive
* Rename to num_bytes and move this calculation outside the loop
[ROCm/composable_kernel commit: 28f68a5a99
2024-01-25 19:53:15 +08:00
Haocong WANG
ec7e5b1331
[GEMM] Optimization for MI200/300. ( #1135 )
...
* Optimize GEMM on MI200/300:
1. Add new blockwise gemm pipeline
2. Add irregular splitk intances
* clang format + typo fix
* Fix a bug
[ROCm/composable_kernel commit: bb63b9732c
2024-01-19 07:02:22 -06:00
Bartłomiej Kocot
d68c1fad45
Revert "[SWDEV-435347] disable instances failed with mainlien compiler ( #1077 )" ( #1101 )
...
This reverts commit a54a4f4281a35e466c86
2024-01-02 11:36:45 +01:00
Artur Wojcik
e9ec2910a0
enable compilation of INSTANCES_ONLY for Windows ( #1082 )
...
* enable compilation of INSTANCES_ONLY for Windows
* suppress ROCMChecks warnings on GoogleTests
* suppress -Wfloat-equal warning on GoogleTests
---------
Co-authored-by: Illia Silin <98187287+illsilin@users.noreply.github.com >
[ROCm/composable_kernel commit: fb5bd51b42
2023-12-20 14:34:53 -08:00
arai713
b5aae16948
Hip tensor permute unit test ( #1068 )
...
* adding files for F32 example
* adding functioning implementation with scalar multiplication and unary operator support
* added fp 16 type check in unary square
* updating scalar multiplication as an operator
* functioning version with scalar operator
* changing strides for col major
* updated column major implementation
* working column major implementation
* cleaned up comments, rearranged/renamed files
* small edits to 3d transpose profiler
* adding test/profiler/instance files for hipTensor permute unit test
* added more test instances
* cleaned up errors, randomized input tensor, added more instances
* turned off time printouts
* removed conflicting transpose profiler
* rearranged some files
[ROCm/composable_kernel commit: 12a8883c48
2023-12-18 21:35:00 -06:00
rocking
53eab49062
layernorm and groupnorm backward data ( #1083 )
...
* rename folder
* Add type string
* Remove typo
* Add deviceOp to backward x
* Add comment to describe the behavior of backward normalization
* Add kernel function, prepare to implement
* implement generic kernel
* Check vector size
* Add sweep once pipeline for small reduce size
* Fix bug of KRaw_ error
* Fix bug of dx stride
* sanity check for mean and rstd
* backward x for groupnorm
* Add bwd x instance
* add layernorm 2d bwd gamma beta instances
* Change save mean var type from f32 to f16 in f16 mode
* Change the example to f16
* Add groupnorm bwd gamma beta instance
* Add groupnorm bwd x instance
* Fix naming
* Add layernorm bwd x ckprofiler
* Add groupnorm bwd x profiler
* clang format
* Rename bwd x to bwd data
* Fix bug of verification in profiler
* Add test of layernorm and groupnorm bwd data
* Add missing cmake
* Add layernorm2d bwd data
* rename fwd example
* Add groupnorm client example
* Fix typo. replace Invarient with Invariant
* Add checking before running the best instance
[ROCm/composable_kernel commit: a69aa2a11a
2023-12-19 04:23:11 +08:00
Bartlomiej Wroblewski
9f2d90a8b6
Optimize fp16 direct load GEMM instances ( #1086 )
...
This PR optimizes fp16 instances of direct load GEMM kernel introduced in #999 and #1052 .
Measured the performance of new instances on CDNA2 GPU and compared it against the performance of the best non-direct-load GEMM instances. Used 76 different GEMM problems.
On average, this change improves the performance of the tested problems by 47%. For cases known as latency-bound, the speedup is around 126%.
[ROCm/composable_kernel commit: ad0a8e4cd2
2023-12-18 11:09:10 +01:00
Illia Silin
a46eb342e3
disabling some fp8 gemm instances to reduce build time ( #1084 )
...
* disabling some fp8 gemm instances to reduce build time
* disable fp8 gemm instances to reduce build time
* remove the unused variable
* build fp8 gemm default and padded instances separately
* fix include pathsc
[ROCm/composable_kernel commit: c004e0d990
2023-12-11 17:49:27 -08:00
Illia Silin
89ca9f73ac
fix clang format ( #1095 )
...
[ROCm/composable_kernel commit: f199035b74
2023-12-08 14:32:37 -08:00
Nicolas Macchioni
25b6fb72e1
Add F8 dtype definition in f16_f8_f16 gemm instances ( #1092 )
...
[ROCm/composable_kernel commit: b4dcd5803f
2023-12-08 13:30:01 -06:00
Bartłomiej Kocot
39b25e2d71
Support broadcast for bias in grouped conv fwd ( #1081 )
...
* Support broadcast for bias in grouped conv fwd
* Fix comment
* Comment fixes
* Remove GK layout
[ROCm/composable_kernel commit: f836984891
2023-12-08 11:07:42 +01:00
Jun Liu
a54a4f4281
[SWDEV-435347] disable instances failed with mainlien compiler ( #1077 )
...
[ROCm/composable_kernel commit: ff24b537cb
2023-12-04 23:45:16 -08:00
Bartlomiej Wroblewski
4f9c5a6996
Add support for double buffering in direct load GEMM kernel ( #1052 )
...
This PR introduces support for double buffering in LDS into GEMM kernels that use direct load instructions.
Direct loads now use inline asm instead of intrinsics. Usage of intrinsics results in compiler adding additional waitcnt instructions what breaks possible load/compute overlap in case of double buffering.
Usage of inline asm results in the need to use sched_barrier in order to make sure that compiler cannot incorrectly reschedule instructions since it does not know the data dependencies between global->LDS and LDS->registers.
[ROCm/composable_kernel commit: bc4bf9bd03
2023-12-03 23:08:47 +01:00
