composable_kernel

mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-14 10:09:41 +00:00

Files

Adam Osewski 0f3d9639a8 Client examples AddFastGelu and FastGelu + instances. (#509 )

* FastGelu support for more data types.

* AddFastGelu & FastGelu instances.

* Client example.

* clang-format

* Remove unused stride variable.

* Add new line at EOF.

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

[ROCm/composable_kernel commit: 43a889b72e]

2022-11-19 22:08:26 -06:00

01_gemm

minor fix in gemm client example (#328 )

2022-07-13 10:54:38 -05:00

02_gemm_add_add_fastgelu

Client examples AddFastGelu and FastGelu + instances. (#509 )

2022-11-19 22:08:26 -06:00

03_gemm_layernorm

Refactor device op implementations into impl subdirectory. (#420 )

2022-10-13 09:05:08 -05:00

04_contraction

Introduce ck::accumulate_n() (#439 )

2022-11-14 19:53:39 -06:00

05_layernorm

Refine layernorm naming and test code (#497 )

2022-11-02 16:57:28 -06:00

06_softmax

Softmax client example (#396 )

2022-09-06 12:22:48 -05:00

07_grouped_conv2d_fwd

Add client example of grouped conv2d forward (data type: fp16) (#488 )

2022-11-09 19:01:58 -06:00

08_fused_attention

Fused attention client example (#494 )

2022-10-27 15:04:31 -06:00

09_quantization

Conv perlayer int8 quantization (#471 )

2022-11-02 13:56:26 -06:00

10_grouped_conv2d_bwd_data

Remove interface 'DeviceGroupedConvBwdData' (#500 )

2022-11-09 18:32:17 -06:00

11_grouped_conv_bwd_weight

Add client example of grouped conv2d backward weight (data type: fp16) (#498 )

2022-11-09 18:50:03 -06:00

12_elementwise_normalization

Add BF16 tests for batched_gemm_softmax_gemm_permute (#504 )

2022-11-15 16:30:23 -06:00

CMakeLists.txt

Replace the obsolete offload-arch flags with GPU_TARGETS and fix a bug. (#437 )

2022-09-22 09:32:25 -05:00

README.md

Improve external interface for GEMM and GEMM+add+add+fastgelu (#311 )

2022-06-30 22:11:00 -05:00

README.md

Client application links to CK library, and therefore CK library needs to be installed before building client applications.

Build

mkdir -p client_example/build
cd client_example/build

cmake                                                                 \
-D CMAKE_CXX_COMPILER=/opt/rocm/bin/hipcc                             \
-D CMAKE_PREFIX_PATH="/opt/rocm;${PATH_TO_CK_INSTALL_DIRECTORY}"      \
..

Build client example

 make -j