Commit Graph

12 Commits

Author SHA1 Message Date
Bartłomiej Kocot
d5ffc2b1b9 Add Grouped Convolution and GEMM documentation (#1719)
* Add Grouped Convolution docs

* Add gemm docs

* Update docs

* fix

[ROCm/composable_kernel commit: 85d6fcd30a]
2025-02-04 16:41:49 +01:00
Haocong WANG
65d6442b4c [GEMM] gemm_universal related optimization (#1453)
* replace buffer_atomic with global_atomic

* fixed global_atomic_add

* added bf16 atomic_add

* format

* clang-format-12

* clean

* clean

* add guards

* Update gtest.cmake

* enabled splitk_gemm_multi_d

* format

* add ckProfiler

* format

* fixed naming

* format

* clean

* clean

* add guards

* fix clang format

* format

* add kbatch printout

* clean

* Add rocm6.2 related gemm optimization

* Limit bf16 atomic usage

* remove redundant RCR gemm_universal instance

* Add RRR fp8 gemm universal instance

* Bug fix

* Add GPU_TARGET guard to FP8/BF8 target

* bug fix

* update cmake

* remove all fp8/bf8 example if arch not support

* Enable fp8 RRR support in ckProfiler

* limit greedy-reverse flag to gemm_universal in ckProfiler

---------

Co-authored-by: Jing Zhang <jizhan@fb.com>
Co-authored-by: Jing Zhang <jizhan@meta.com>
Co-authored-by: zjing14 <zhangjing14@gmail.com>
Co-authored-by: Illia Silin <98187287+illsilin@users.noreply.github.com>
Co-authored-by: illsilin <Illia.Silin@amd.com>

[ROCm/composable_kernel commit: 3049b5467c]
2024-08-14 10:42:30 +08:00
Bartłomiej Kocot
68b2757f11 Add two stage grouped conv bwd weight kernel (#1280)
[ROCm/composable_kernel commit: 0b6b5d1785]
2024-05-08 09:53:24 +02:00
amoskvic
e9d9e4d0ad Style improvement: improving type alias usage consistency in gemm-related client examples. Also copyright year update for all client examples. (#1180)
Co-authored-by: Arseny Moskvichev <amoskvic@amd.com>

[ROCm/composable_kernel commit: a776978cbe]
2024-02-28 16:39:03 -08:00
Illia Silin
2e8781c151 Split the static library into several files. (#1044)
* spolit the static library into several

* update lib paths and fix client example

* do not use device_mha_operarions for client examples

* use appropriate libs to link to client examples

* remove the gpu/transpose path from the list

* try fixing clinet examples 3,4,9

* add necessary libs for client examples

* fix the layernorm client example

* fix the client examples 23 and 24

* fix typo

* add interface library and refresh clang format

[ROCm/composable_kernel commit: 7965d66a81]
2023-11-28 11:17:37 -08:00
Rostyslav Geyyer
9866e150da Add conv bwd weight client example (#1005)
* Add conv bwd weight client example

* Update instance selector

* Fake the conversion

* Bring the conversion back

[ROCm/composable_kernel commit: 5356c4a943]
2023-11-13 11:16:04 -06:00
Bartłomiej Kocot
33496a08cc Add wei_strides to grouped conv3d wei to keep consistency (#817)
* Add wei_strides to grouped conv3d wei to keep consistency

* Fix strides in client examples

* Unify backward weight api with forward

* Fix for example

* Fixes for examples

---------

Co-authored-by: zjing14 <zhangjing14@gmail.com>

[ROCm/composable_kernel commit: 22443f7aae]
2023-08-07 10:23:45 -05:00
Bartłomiej Kocot
dd6d0dd3b9 Support NHWGC conv2d_bwd_weight (#769)
* Support NHWGC conv2d_bwd_weight

* Fix client example

* Fix client example

* Fix comments

* Redesign grouped_conv_bwd_weight instances

* Clang format fix

---------

Co-authored-by: zjing14 <zhangjing14@gmail.com>

[ROCm/composable_kernel commit: 1ee99dcaa6]
2023-07-12 08:25:02 -05:00
Illia Silin
d40b8d5e2c update copyright headers (#726)
[ROCm/composable_kernel commit: b94fd0b227]
2023-05-31 18:46:57 -05:00
ltqin
ae95445d01 Grouped conv1d client example (#589)
* add conv1d fwd client example

* change 07_grouped_conv2d_fwd to 07_grouped_convnd_fwd

* add conv1d bwd weight

---------

Co-authored-by: zjing14 <zhangjing14@gmail.com>

[ROCm/composable_kernel commit: 830d37a7d5]
2023-02-22 11:55:21 -06:00
Adam Osewski
85acf7ac2f Conv3D FWD BWD WRW fp16 fp32 client examples (#559)
* Conv3d bwd weight client example.

* Update year in license

* Convolution bwd data 3D fp16/fp32 client example.

* Client example for convnd fwd fp16 fp32

* clang-format

* Review remarks.

* Fix compiler err.

* Update data layout to standard one.

* Add conv 3d fwd NDHWGC instances

* clang-format

* Conv3d fwd NDHWGC instances.

---------

Co-authored-by: Adam Osewski <aosewski@amd.com>
Co-authored-by: zjing14 <zhangjing14@gmail.com>

[ROCm/composable_kernel commit: e9fd122889]
2023-02-15 11:16:47 -06:00
Po Yen Chen
93f036f2c3 Add client example of grouped conv2d backward weight (data type: fp16) (#498)
* Remove redundant CMake setting

* Extract common code from files

* Rename folder 'convnd' to 'conv'

* Use std::array<> to accept compile-time kwnown # of arguments

* Fix compilation error of tuning parameter

* In example, use same setting as unit-test

* Remove no-longer used include directive

* Add interface for grouped conv bwd weight

* Add group support for conv bwd weight

* Add grouped conv bwd weight example

* Use group parameter in example

* Rename example folder

* Remove non-grouped version example source files

* Rename device op template

* Add group support to convolution backward weight

* Remove debug messages

* Use smaller group size in example

* Use named variable as loop terminate condition

* Prettify example output message

* Enlarge used grid size

* Allow real grid size exceeds expected grid size

* Rename interface file

* Add client example for grouped conv2d bwd weight

* Fix wrong include directive

* Rename client example folder

[ROCm/composable_kernel commit: 38470e0497]
2022-11-09 18:50:03 -06:00