ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-17 19:40:04 +00:00
Code Issues Packages Projects Releases Wiki Activity
Files
a052bfa81a827ddd770a40fc08e0edc7e2c07983
composable_kernel/include/ck/tensor_operation/gpu/grid
History
Bartłomiej Kocot 48034577d4 Grouped Convolution Backward Weight Explicit GEMM (#2282) 
* Grouped conv bwd weight explicit gemm

* 3d

* cmake fixes

* fix test

* fix

[ROCm/composable_kernel commit: 050cad09b5]
2025-06-06 10:30:08 +02:00
..
batchnorm_multiblock
…
gemm_layernorm
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
normalization
…
block_to_ctile_map.hpp
MI308 fix for streamk 1-Tile floating point exception (#2101)
2025-04-21 11:44:07 -07:00
gridwise_2d_multiple_reduction_multiblock.hpp
…
gridwise_2d_multiple_reduction_threadwise.hpp
…
gridwise_2d_reduction_multiblock.hpp
…
gridwise_2d_reduction_threadwise_multi_d.hpp
…
gridwise_2d_reduction_threadwise.hpp
…
gridwise_batched_gemm_gemm_xdl_cshuffle_v1.hpp
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
gridwise_batched_gemm_multiple_d_gemm_multiple_d_xdl_cshuffle_v1.hpp
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
gridwise_batched_gemm_multiple_d_softmax_gemm_xdl_cshuffle_v1.hpp
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
gridwise_batched_gemm_softmax_gemm_wmma_cshuffle.hpp
…
gridwise_batched_gemm_softmax_gemm_xdl_cshuffle_v1.hpp
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
gridwise_batchnorm_backward_blockwise_welford.hpp
…
gridwise_batchnorm_forward_blockwise_welford.hpp
…
gridwise_elementwise_1d_scale.hpp
…
gridwise_elementwise_2d.hpp
Grouped conv backward data GKCYX support (#2029)
2025-04-01 13:24:38 -07:00
gridwise_elementwise_layernorm_welford_variance.hpp
…
gridwise_fpAintB_gemm_wmma.hpp
…
gridwise_gemm_bias_add_reduce_xdl_cshuffle_v1.hpp
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
gridwise_gemm_dl_multiple_d.hpp
…
gridwise_gemm_dl_v1r3.hpp
…
gridwise_gemm_dpp.hpp
…
gridwise_gemm_multiple_abd_xdl_cshuffle.hpp
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
gridwise_gemm_multiple_d_multiple_r_xdl_cshuffle.hpp
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
gridwise_gemm_multiple_d_wmma_cshuffle.hpp
…
gridwise_gemm_multiple_d_xdl_cshuffle_lds_direct_load.hpp
Optimized GEMMs for MX FP4/8 (#2294)
2025-06-05 13:54:15 -06:00
gridwise_gemm_multiple_d_xdl_cshuffle.hpp
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
gridwise_gemm_multiple_d_xdl_splitk_cshuffle.hpp
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
gridwise_gemm_pipeline_selector.hpp
…
gridwise_gemm_pipeline_v1.hpp
…
gridwise_gemm_pipeline_v2.hpp
…
gridwise_gemm_pipeline_v3.hpp
…
gridwise_gemm_pipeline_v4_direct_load.hpp
…
gridwise_gemm_reduce_xdl_cshuffle_v1.hpp
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
gridwise_gemm_split_k_multiple_d_xdl_cshuffle_v2.hpp
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
gridwise_gemm_split_k_multiple_d_xdl_cshuffle.hpp
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
gridwise_gemm_waveletmodel.hpp
…
gridwise_gemm_wmma_cshuffle_v3.hpp
WMMA GEMM universal pipeline v1, mixed precision and paddings, examples (#2230)
2025-06-04 12:22:33 +06:00
gridwise_gemm_wmma.hpp
…
gridwise_gemm_xdl_cshuffle_conv_v3.hpp
Grouped Convolution Backward Weight Explicit GEMM (#2282)
2025-06-06 10:30:08 +02:00
gridwise_gemm_xdl_cshuffle_streamk_v3.hpp
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
gridwise_gemm_xdl_cshuffle_v1.hpp
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
gridwise_gemm_xdl_cshuffle_v2.hpp
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
gridwise_gemm_xdl_cshuffle_v3_b_preshuffle.hpp
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
gridwise_gemm_xdl_cshuffle_v3_b_scale.hpp
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
gridwise_gemm_xdl_cshuffle_v3_multi_abd.hpp
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
gridwise_gemm_xdl_cshuffle_v3_multi_d_ab_scale.hpp
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
gridwise_gemm_xdl_cshuffle_v3_multi_d_b_preshuffle.hpp
Optimized GEMMs for MX FP4/8 (#2294)
2025-06-05 13:54:15 -06:00
gridwise_gemm_xdl_cshuffle_v3_multi_d.hpp
Grouped Convolution Backward Weight Explicit GEMM (#2282)
2025-06-06 10:30:08 +02:00
gridwise_gemm_xdl_cshuffle_v3_mx_bpreshuffle.hpp
Optimized GEMMs for MX FP4/8 (#2294)
2025-06-05 13:54:15 -06:00
gridwise_gemm_xdl_cshuffle_v3_mx.hpp
Optimized GEMMs for MX FP4/8 (#2294)
2025-06-05 13:54:15 -06:00
gridwise_gemm_xdl_cshuffle_v3.hpp
Optimized GEMMs for MX FP4/8 (#2294)
2025-06-05 13:54:15 -06:00
gridwise_gemm_xdl_layernorm_cshuffle_v1.hpp
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
gridwise_gemm_xdl_waveletmodel_cshuffle.hpp
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
gridwise_gemm_xdlops_bwd_weight.hpp
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
gridwise_gemm_xdlops_skip_b_lds_v1.hpp
…
gridwise_gemm_xdlops_splitk_lds_direct_load.hpp
Optimized GEMMs for MX FP4/8 (#2294)
2025-06-05 13:54:15 -06:00
gridwise_gemm_xdlops_streamk.hpp
…
gridwise_gemm_xdlops_v2r3.hpp
…
gridwise_gemm_xdlops_v2r4.hpp
…
gridwise_gemm_xdlops_v2r4r2.hpp
Split env.hpp header from the ck.hpp header. (#2049)
2025-04-03 15:30:21 -07:00
gridwise_gemm_xdlops_v3r1.hpp
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
gridwise_gemm_xdlops_v3r2.hpp
…
gridwise_gemm_xdlops_v3r3.hpp
…
gridwise_moe_gemm.hpp
Use new mfma instructions for FP8 on gfx950 (#2202)
2025-05-19 17:29:51 -07:00
gridwise_permute.hpp
…
gridwise_put_element_1d.hpp
…
gridwise_set_buffer_value.hpp
…
gridwise_set_multiple_buffer_value.hpp
…
gridwise_softmax.hpp
…
gridwise_sparse_embeddings_forward_layernorm_builtins.hpp
Replace buffer load/store intrinsics with builtins (#1876)
2025-03-05 14:33:28 -08:00
gridwise_sparse_embeddings_forward_layernorm.hpp
…
gridwise_tensor_rearrange.hpp
…