Adam Osewski 8bb2bb4a05 Grouped Gemm + SplitK + simplified Kernel Args (#669) ...

* simplify karg in device/grid split-k op

* fix mk_kn_mn instances

* add more instances

* B2C with 3D grid for KSplit

* Remove unused code.

* Use default B2C (3D grid) in grid gemm v2r4r2.

* Device gemm splitk use B2C map.

* Device GroupedGemmXdlSplitKCShuffle

* Example for GroupedGemm Xdl SplitK

* Introduce Device GroupedGemmSplitK

* Fix updating kbatch size.

* Add instance mk-nk-mn

* Enable set kbatch in profiler.

* Add GGemmSplitK mk-kn-mn instances

* Add more instances & split into multiple files.

* minor fix

* tuning

* clean

* disabled failed instances

* use pipe v2

* Ignore arg on not supported arch.

* fix warning

---------

Co-authored-by: carlushuang <carlus.huang@amd.com>
Co-authored-by: Adam Osewski <aosewski@amd.com>
Co-authored-by: zjing14 <zhangjing14@gmail.com>
Co-authored-by: Jing Zhang <jizhan@amd.com>
Co-authored-by: root <root@ctr-ubbsmc15.amd.com>

2023-04-24 15:43:36 -05:00

..

01_gemm

[gfx110x] support Navi3x architectures. (#628)

2023-03-09 07:56:40 -06:00

02_gemm_bilinear

[gfx110x] support Navi3x architectures. (#628)

2023-03-09 07:56:40 -06:00

03_gemm_bias_relu

Rangify constructor of HostTensorDescriptor & Tensor<> (#445)

2022-11-11 11:36:01 -06:00

04_gemm_add_add_fastgelu

Fast GeLU using built-in function (#587)

2023-02-26 23:19:11 -06:00

09_convnd_fwd

Add multiple d gridwise gemm on Navi21 for ResNet50 (#517)

2022-12-02 11:42:31 -06:00

10_convnd_fwd_multiple_d_multiple_reduce

Rangify constructor of HostTensorDescriptor & Tensor<> (#445)

2022-11-11 11:36:01 -06:00

12_reduce

Use double for all scaling values and float-point constant values at the Device Op API (#557)

2023-01-18 12:02:50 -06:00

13_pool2d_fwd

Rangify constructor of HostTensorDescriptor & Tensor<> (#445)

2022-11-11 11:36:01 -06:00

14_gemm_quantization

gemm/Conv xdlops + dlops quantization (#625)

2023-03-15 15:29:40 -05:00

15_grouped_gemm

Grouped Gemm + SplitK + simplified Kernel Args (#669)

2023-04-24 15:43:36 -05:00

16_gemm_multi_d_multi_reduces

Rangify constructor of HostTensorDescriptor & Tensor<> (#445)

2022-11-11 11:36:01 -06:00

17_convnd_bwd_data

Add Conv Backward Data on Navi21 for ResNet50 (#499)

2022-11-15 16:22:20 -06:00

18_batched_gemm_reduce

Rangify constructor of HostTensorDescriptor & Tensor<> (#445)

2022-11-11 11:36:01 -06:00

19_binary_elementwise

Batchnorm inference instances, external API, client examples and gtests (#531)

2023-01-25 17:09:04 -06:00

20_grouped_conv_bwd_weight

[Navi3x Bug Fix] fix typo to accept MNKPadding flag correctly. (#597)

2023-03-01 12:07:42 -06:00

21_gemm_layernorm

Gemm+layernorm instance, ckProfiler, client example (#568)

2023-02-09 15:02:55 -06:00

22_cgemm

Rangify constructor of HostTensorDescriptor & Tensor<> (#445)

2022-11-11 11:36:01 -06:00

23_softmax

Use double for all scaling values and float-point constant values at the Device Op API (#557)

2023-01-18 12:02:50 -06:00

24_batched_gemm

Rangify constructor of HostTensorDescriptor & Tensor<> (#445)

2022-11-11 11:36:01 -06:00

25_gemm_bias_e_permute

Introduce ck::accumulate_n() (#439)

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

26_contraction

Add contraction_fp64 example (#570)

2023-02-15 12:00:58 -06:00

27_layernorm

Improve normalization (#580)

2023-02-15 11:59:35 -06:00

28_grouped_gemm_bias_e_permute

Introduce ck::accumulate_n() (#439)

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

29_batched_gemm_bias_e_permute

[gfx110x] support Navi3x architectures. (#628)

2023-03-09 07:56:40 -06:00

30_grouped_conv_fwd_multiple_d

Add a denorm test fix (#603)

2023-03-29 15:05:32 -05:00

31_batched_gemm_gemm

Rangify constructor of HostTensorDescriptor & Tensor<> (#445)

2022-11-11 11:36:01 -06:00

32_batched_gemm_scale_softmax_gemm

Add BF16 tests for batched_gemm_softmax_gemm_permute (#504)

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

33_multiple_reduce

Use double for all scaling values and float-point constant values at the Device Op API (#557)

2023-01-18 12:02:50 -06:00

34_batchnorm

Batchnorm inference instances, external API, client examples and gtests (#531)

2023-01-25 17:09:04 -06:00

35_splitK_gemm

Rangify constructor of HostTensorDescriptor & Tensor<> (#445)

2022-11-11 11:36:01 -06:00

36_sparse_embedding

add multi embeddings support (#542)

2023-01-18 11:32:12 -06:00

37_batched_gemm_add_add_relu_gemm_add

Rangify constructor of HostTensorDescriptor & Tensor<> (#445)

2022-11-11 11:36:01 -06:00

38_grouped_conv_bwd_data_multiple_d

Rangify STL algorithms (#438)

2022-11-14 15:17:28 -06:00

39_permute

Rangify constructor of HostTensorDescriptor & Tensor<> (#445)

2022-11-11 11:36:01 -06:00

40_conv2d_fwd_quantization

Revise layout of group convolution (#675)

2023-04-23 23:40:00 -05:00

41_grouped_conv_conv_fwd

Introduce ck::accumulate_n() (#439)

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

42_groupnorm

Groupnorm + swish external api (#668)

2023-04-10 08:02:17 -05:00

43_splitk_gemm_bias_e_permute

Example contraction splitk (#430)

2022-10-11 17:54:34 -05:00

44_elementwise_permute

Batchnorm inference instances, external API, client examples and gtests (#531)

2023-01-25 17:09:04 -06:00

45_elementwise_normalization

Fused elementwise normalization (#492)

2022-11-03 12:01:58 -06:00

46_gemm_add_multiply

Conv3D FWD BWD WRW fp16 fp32 client examples (#559)

2023-02-15 11:16:47 -06:00

47_gemm_bias_softmax_gemm_permute

Add GemmAddSoftmaxGemm support for MSFT ORT (instances and client API) (#576)

2023-02-08 14:34:45 -06:00

CMakeLists.txt

Add packages for examples and profiler (#502)

2022-11-10 13:19:33 -06:00