composable_kernel-devel-7.14.0-2.fc46 in Fedora Rawhide
↵ Return to the main page of composable_kernel-devel
View build
Search for updates
Package Info (Data from x86_64 build)
🠗 Changelog
🠗 Dependencies
🠗 Provides
🠗 Files
Changelog
| Date | Author | Change |
|---|---|---|
| 2026-08-15 | Tom Rix <Tom dot Rix at amd dot com> - 7.14.0-2 | - Fix files |
| 2026-08-13 | Tom Rix <Tom dot Rix at amd dot com> - 7.14.0-1 | - Update to 7.14 |
| 2026-07-30 | Tom Rix <Tom dot Rix at amd dot com> - 7.2.0-5 | - Add --with preview |
| 2026-07-15 | Fedora Release Engineering <releng at fedoraproject dot org> - 7.2.0-4 | - Rebuilt for https://fedoraproject.org/wiki/Fedora_45_Mass_Rebuild |
| 2026-06-20 | Tom Rix <Tom dot Rix at amd dot com> - 7.2.0-3 | - merge compat changes |
| 2026-03-07 | Tom Rix <Tom dot Rix at amd dot com> - 7.2.0-2 | - Build what is needed for miopen |
| 2026-02-08 | Tom Rix <Tom dot Rix at amd dot com> - 7.2.0-1 | - Update to 7.2.0 |
| 2026-01-29 | Tom Rix <Tom dot Rix at amd dot com> - 7.1.1-7 | - Disable gpu library building |
| 2026-01-16 | Fedora Release Engineering <releng at fedoraproject dot org> - 7.1.1-6 | - Rebuilt for https://fedoraproject.org/wiki/Fedora_44_Mass_Rebuild |
| 2026-01-16 | Fedora Release Engineering <releng at fedoraproject dot org> - 7.1.1-5 | - Rebuilt for https://fedoraproject.org/wiki/Fedora_44_Mass_Rebuild |
Dependencies
- cmake-filesystem
- composable_kernel = 7.14.0-2.fc46
- composable_kernel7.2
- rocm-filesystem
Provides
- cmake(composable_kernel)
- composable_kernel-devel
- composable_kernel-devel(x86-64)
- composable_kernel-static
Files
- usr/
- include/
- ck/
- BUILD_TIME_OPTIMIZATION.md
- README.md
- ck.hpp
- config.h
- config.h.in
- filesystem.hpp
- host_utility/
- device_prop.hpp
- flush_cache.hpp
- hip_check_error.hpp
- io.hpp
- kernel_launch.hpp
- stream_utility.hpp
- library/
- reference_tensor_operation/
- cpu/
- reference_avgpool_bwd.hpp
- reference_batched_gemm.hpp
- reference_batchnorm_backward.hpp
- reference_batchnorm_forward.hpp
- reference_batchnorm_infer.hpp
- reference_cgemm.hpp
- reference_column_to_image.hpp
- reference_contraction.hpp
- reference_conv_bwd_data.hpp
- reference_conv_bwd_weight.hpp
- reference_conv_fwd.hpp
- reference_conv_fwd_bias_activation.hpp
- reference_conv_fwd_bias_activation_add.hpp
- reference_elementwise.hpp
- reference_fpAintB_gemm.hpp
- reference_gemm.hpp
- reference_gemm_layernorm.hpp
- reference_gemm_multi_abd.hpp
- reference_gemm_multiple_d.hpp
- reference_groupnorm.hpp
- reference_groupnorm_bwd.hpp
- reference_image_to_column.hpp
- reference_layernorm.hpp
- reference_layernorm_bwd.hpp
- reference_maxpool_bwd.hpp
- reference_moe_gemm.hpp
- reference_moe_gemm1_blockscale.hpp
- reference_moe_gemm1_blockscale_splitk.hpp
- reference_moe_gemm2.hpp
- reference_moe_gemm2_blockscale.hpp
- reference_moe_mx_gemm1.hpp
- reference_moe_mx_gemm2.hpp
- reference_mx_gemm.hpp
- reference_pool_fwd.hpp
- reference_reduce.hpp
- reference_softmax.hpp
- reference_sparse_embedding3_forward_layernorm.hpp
- gpu/
- naive_conv_bwd_data_gpu.hpp
- naive_conv_bwd_weight_gpu.hpp
- naive_conv_fwd_gpu.hpp
- naive_conv_utils.hpp
- reference_gemm.hpp
- cpu/
- tensor_operation_instance/
- add_device_operation_instance.hpp
- add_grouped_conv_bwd_wei_exp_device_operation_instance.hpp
- device_operation_instance_factory.hpp
- gpu/
- avg_pool2d_bwd.hpp
- avg_pool3d_bwd.hpp
- batched_gemm.hpp
- batched_gemm_add_relu_gemm_add.hpp
- batched_gemm_b_scale.hpp
- batched_gemm_bias_permute.hpp
- batched_gemm_bias_softmax_gemm_permute.hpp
- batched_gemm_gemm.hpp
- batched_gemm_multi_d.hpp
- batched_gemm_softmax_gemm.hpp
- batched_gemm_softmax_gemm_permute.hpp
- batchnorm_backward.hpp
- batchnorm_forward.hpp
- batchnorm_infer.hpp
- contraction/
- device_contraction_instance.hpp
- contraction_bilinear.hpp
- contraction_scale.hpp
- conv_tensor_rearrange/
- device_column_to_image_instance.hpp
- device_image_to_column_instance.hpp
- conv_tensor_rearrange.hpp
- convolution_backward_data.hpp
- convolution_forward.hpp
- device_elementwise_instance.hpp
- device_gemm_mean_squaremean_instance.hpp
- device_gemm_xdl_c_shuffle_fp8_fp8_fp8_mk_kn_mn_v1_instance.hpp
- device_gemm_xdl_c_shuffle_fp8_fp8_fp8_mk_kn_mn_v1_interwave_instance.hpp
- device_gemm_xdl_c_shuffle_fp8_fp8_fp8_mk_kn_mn_v2_instance.hpp
- elementwise_normalization.hpp
- gemm.hpp
- gemm_ab_scale.hpp
- gemm_add.hpp
- gemm_add_add_fastgelu.hpp
- gemm_add_fastgelu.hpp
- gemm_add_multiply.hpp
- gemm_add_relu.hpp
- gemm_add_relu_add_layernorm.hpp
- gemm_add_silu.hpp
- gemm_b_scale.hpp
- gemm_bilinear.hpp
- gemm_blockscale_wp.hpp
- gemm_dl.inc
- gemm_dpp.inc
- gemm_fastgelu.hpp
- gemm_multi_abd.hpp
- gemm_multiply_add.hpp
- gemm_multiply_multiply.hpp
- gemm_multiply_multiply_wp.hpp
- gemm_mx.hpp
- gemm_splitk.hpp
- gemm_universal.hpp
- gemm_universal_batched.hpp
- gemm_universal_preshuffle.hpp
- gemm_universal_preshuffle.inc
- gemm_universal_reduce.hpp
- gemm_universal_wmma.inc
- gemm_universal_xdl.inc
- gemm_wmma.inc
- gemm_xdl.inc
- grouped_conv_bwd_data/
- device_grouped_conv_bwd_data_transpose_xdl_instance.hpp
- device_grouped_conv_bwd_data_wmma_f16_instance.hpp
- device_grouped_conv_bwd_data_wmma_i8_instance.hpp
- device_grouped_conv_bwd_data_wmma_v3_bilinear_instance.hpp
- device_grouped_conv_bwd_data_wmma_v3_instances.hpp
- device_grouped_conv_bwd_data_wmma_v3_scale_instance.hpp
- device_grouped_conv_bwd_data_xdl_bilinear_instance.hpp
- device_grouped_conv_bwd_data_xdl_instance.hpp
- device_grouped_conv_bwd_data_xdl_scale_instance.hpp
- device_grouped_conv_bwd_data_xdl_v3_instance.hpp
- grouped_conv_bwd_weight/
- device_exp_gemm_wmma_universal_km_kn_mn_instance.hpp
- device_exp_gemm_xdl_universal_km_kn_mn_instance.hpp
- device_grouped_conv_bwd_weight_dl_instance.hpp
- device_grouped_conv_bwd_weight_two_stage_wmma_instance.hpp
- device_grouped_conv_bwd_weight_two_stage_xdl_instance.hpp
- device_grouped_conv_bwd_weight_v3_wmma_instance.hpp
- device_grouped_conv_bwd_weight_v3_xdl_instance.hpp
- device_grouped_conv_bwd_weight_wavelet_xdl_instance.hpp
- device_grouped_conv_bwd_weight_wmma_bilinear_instance.hpp
- device_grouped_conv_bwd_weight_wmma_scale_instance.hpp
- device_grouped_conv_bwd_weight_xdl_bilinear_instance.hpp
- device_grouped_conv_bwd_weight_xdl_instance.hpp
- device_grouped_conv_bwd_weight_xdl_scale_instance.hpp
- grouped_conv_fwd/
- device_grouped_conv_fwd_dl_instance.hpp
- device_grouped_conv_fwd_wmma_cshufflev3_bilinear_instance.hpp
- device_grouped_conv_fwd_wmma_cshufflev3_binary_outelementop_instance.hpp
- device_grouped_conv_fwd_wmma_cshufflev3_dynamic_op_instance.hpp
- device_grouped_conv_fwd_wmma_cshufflev3_instance.hpp
- device_grouped_conv_fwd_wmma_cshufflev3_large_tensor_instance.hpp
- device_grouped_conv_fwd_wmma_cshufflev3_outelementop_instance.hpp
- device_grouped_conv_fwd_wmma_cshufflev3_scale_instance.hpp
- device_grouped_conv_fwd_wmma_cshufflev3_scaleadd_ab_instance.hpp
- device_grouped_conv_fwd_wmma_cshufflev3_scaleadd_scaleadd_relu_instance.hpp
- device_grouped_conv_fwd_wmma_cshufflev3_wave_transfer_instance.hpp
- device_grouped_conv_fwd_wmma_instance.hpp
- device_grouped_conv_fwd_xdl_bilinear_instance.hpp
- device_grouped_conv_fwd_xdl_binary_outelementop_instance.hpp
- device_grouped_conv_fwd_xdl_comp_instance.hpp
- device_grouped_conv_fwd_xdl_dynamic_op_instance.hpp
- device_grouped_conv_fwd_xdl_instance.hpp
- device_grouped_conv_fwd_xdl_large_tensor_instance.hpp
- device_grouped_conv_fwd_xdl_mem_instance.hpp
- device_grouped_conv_fwd_xdl_merged_groups_instance.hpp
- device_grouped_conv_fwd_xdl_outelementop_instance.hpp
- device_grouped_conv_fwd_xdl_scale_instance.hpp
- device_grouped_conv_fwd_xdl_scaleadd_ab_instance.hpp
- device_grouped_conv_fwd_xdl_scaleadd_scaleadd_relu_instance.hpp
- grouped_convolution_backward_data.hpp
- grouped_convolution_backward_data_bilinear.hpp
- grouped_convolution_backward_data_scale.hpp
- grouped_convolution_backward_data_wmma.inc
- grouped_convolution_backward_data_xdl.inc
- grouped_convolution_backward_weight.hpp
- grouped_convolution_backward_weight_bilinear.hpp
- grouped_convolution_backward_weight_dl.inc
- grouped_convolution_backward_weight_explicit_wmma.inc
- grouped_convolution_backward_weight_explicit_xdl.inc
- grouped_convolution_backward_weight_scale.hpp
- grouped_convolution_backward_weight_wmma.inc
- grouped_convolution_backward_weight_xdl.inc
- grouped_convolution_forward.hpp
- grouped_convolution_forward_bias_bnorm_clamp.hpp
- grouped_convolution_forward_bias_bnorm_clamp_wmma_cshufflev3.inc
- grouped_convolution_forward_bias_bnorm_clamp_xdl.inc
- grouped_convolution_forward_bias_clamp.hpp
- grouped_convolution_forward_bias_clamp_wmma_cshufflev3.inc
- grouped_convolution_forward_bias_clamp_xdl.inc
- grouped_convolution_forward_bilinear.hpp
- grouped_convolution_forward_clamp.hpp
- grouped_convolution_forward_clamp_wmma_cshufflev3.inc
- grouped_convolution_forward_clamp_xdl.inc
- grouped_convolution_forward_comp_xdl.inc
- grouped_convolution_forward_convinvscale.hpp
- grouped_convolution_forward_convscale.hpp
- grouped_convolution_forward_convscale_add.hpp
- grouped_convolution_forward_convscale_relu.hpp
- grouped_convolution_forward_dl.inc
- grouped_convolution_forward_dynamic_op.hpp
- grouped_convolution_forward_mem_inter_wmma_cshufflev3.inc
- grouped_convolution_forward_mem_inter_xdl.inc
- grouped_convolution_forward_mem_intra_wmma_cshufflev3.inc
- grouped_convolution_forward_mem_intra_xdl.inc
- grouped_convolution_forward_scale.hpp
- grouped_convolution_forward_scaleadd_ab.hpp
- grouped_convolution_forward_scaleadd_scaleadd_relu.hpp
- grouped_convolution_forward_wmma.inc
- grouped_convolution_forward_wmma_cshufflev3.inc
- grouped_convolution_forward_wmma_large_tensor.inc
- grouped_convolution_forward_xdl.inc
- grouped_convolution_forward_xdl_large_tensor.inc
- grouped_convolution_forward_xdl_merged_groups.inc
- grouped_gemm/
- device_grouped_gemm_wmma_fixed_nk_instance.hpp
- device_grouped_gemm_wmma_splitk_instance.hpp
- device_grouped_gemm_xdl_splitk_instance.hpp
- grouped_gemm.hpp
- grouped_gemm_bias.hpp
- grouped_gemm_fastgelu.hpp
- grouped_gemm_fixed_nk.hpp
- grouped_gemm_multi_abd_fixed_nk.hpp
- grouped_gemm_tile_loop/
- device_grouped_gemm_tile_loop_multiply_wmma_instance.hpp
- device_grouped_gemm_tile_loop_wmma_instance.hpp
- grouped_gemm_tile_loop.hpp
- grouped_gemm_tile_loop_multiply.hpp
- groupnorm_bwd_data.hpp
- groupnorm_bwd_gamma_beta.hpp
- layernorm_bwd_data.hpp
- layernorm_bwd_gamma_beta.hpp
- max_pool_bwd.hpp
- normalization_fwd.hpp
- normalization_fwd_swish.hpp
- permute_scale/
- device_permute_scale_instances.hpp
- permute_scale.hpp
- pool2d_fwd.hpp
- pool3d_fwd.hpp
- quantization/
- gemm_quantization.hpp
- grouped_convolution_bias_forward_perchannel_quantization.hpp
- grouped_convolution_bias_forward_perlayer_quantization.hpp
- grouped_convolution_forward_perchannel_quantization.hpp
- grouped_convolution_forward_perlayer_quantization.hpp
- reduce/
- device_reduce_instance.hpp
- device_reduce_instance_blockwise.hpp
- device_reduce_instance_blockwise_b16_f32_b16_add.hpp
- device_reduce_instance_blockwise_b16_f32_b16_amax.hpp
- device_reduce_instance_blockwise_b16_f32_b16_avg.hpp
- device_reduce_instance_blockwise_b16_f32_b16_max.hpp
- device_reduce_instance_blockwise_b16_f32_b16_min.hpp
- device_reduce_instance_blockwise_b16_f32_b16_norm2.hpp
- device_reduce_instance_blockwise_f16_f16_f16_amax.hpp
- device_reduce_instance_blockwise_f16_f16_f16_max.hpp
- device_reduce_instance_blockwise_f16_f16_f16_min.hpp
- device_reduce_instance_blockwise_f16_f32_f16_add.hpp
- device_reduce_instance_blockwise_f16_f32_f16_avg.hpp
- device_reduce_instance_blockwise_f16_f32_f16_norm2.hpp
- device_reduce_instance_blockwise_f32_f32_f32_add.hpp
- device_reduce_instance_blockwise_f32_f32_f32_amax.hpp
- device_reduce_instance_blockwise_f32_f32_f32_avg.hpp
- device_reduce_instance_blockwise_f32_f32_f32_max.hpp
- device_reduce_instance_blockwise_f32_f32_f32_min.hpp
- device_reduce_instance_blockwise_f32_f32_f32_norm2.hpp
- device_reduce_instance_blockwise_f32_f64_f32_add.hpp
- device_reduce_instance_blockwise_f32_f64_f32_avg.hpp
- device_reduce_instance_blockwise_f32_f64_f32_norm2.hpp
- device_reduce_instance_blockwise_f64_f64_f64_add.hpp
- device_reduce_instance_blockwise_f64_f64_f64_amax.hpp
- device_reduce_instance_blockwise_f64_f64_f64_avg.hpp
- device_reduce_instance_blockwise_f64_f64_f64_max.hpp
- device_reduce_instance_blockwise_f64_f64_f64_min.hpp
- device_reduce_instance_blockwise_f64_f64_f64_norm2.hpp
- device_reduce_instance_blockwise_i8_i32_i8_add.hpp
- device_reduce_instance_blockwise_i8_i32_i8_avg.hpp
- device_reduce_instance_blockwise_i8_i8_i8_amax.hpp
- device_reduce_instance_blockwise_i8_i8_i8_max.hpp
- device_reduce_instance_blockwise_i8_i8_i8_min.hpp
- device_reduce_instance_impl_common.hpp
- device_reduce_instance_multiblock_atomic_add.hpp
- device_reduce_instance_multiblock_atomic_add_b16_f32_f32_add.hpp
- device_reduce_instance_multiblock_atomic_add_b16_f32_f32_avg.hpp
- device_reduce_instance_multiblock_atomic_add_f16_f32_f32_add.hpp
- device_reduce_instance_multiblock_atomic_add_f16_f32_f32_avg.hpp
- device_reduce_instance_multiblock_atomic_add_f32_f32_f32_add.hpp
- device_reduce_instance_multiblock_atomic_add_f32_f32_f32_avg.hpp
- device_reduce_instance_multiblock_atomic_add_f32_f64_f32_add.hpp
- device_reduce_instance_multiblock_atomic_add_f32_f64_f32_avg.hpp
- device_reduce_instance_multiblock_atomic_add_f64_f64_f64_add.hpp
- device_reduce_instance_multiblock_atomic_add_f64_f64_f64_avg.hpp
- device_reduce_instance_threadwise.hpp
- device_reduce_instance_threadwise_b16_f32_b16_add.hpp
- device_reduce_instance_threadwise_b16_f32_b16_amax.hpp
- device_reduce_instance_threadwise_b16_f32_b16_avg.hpp
- device_reduce_instance_threadwise_b16_f32_b16_max.hpp
- device_reduce_instance_threadwise_b16_f32_b16_min.hpp
- device_reduce_instance_threadwise_b16_f32_b16_norm2.hpp
- device_reduce_instance_threadwise_f16_f16_f16_amax.hpp
- device_reduce_instance_threadwise_f16_f16_f16_max.hpp
- device_reduce_instance_threadwise_f16_f16_f16_min.hpp
- device_reduce_instance_threadwise_f16_f32_f16_add.hpp
- device_reduce_instance_threadwise_f16_f32_f16_avg.hpp
- device_reduce_instance_threadwise_f16_f32_f16_norm2.hpp
- device_reduce_instance_threadwise_f32_f32_f32_add.hpp
- device_reduce_instance_threadwise_f32_f32_f32_amax.hpp
- device_reduce_instance_threadwise_f32_f32_f32_avg.hpp
- device_reduce_instance_threadwise_f32_f32_f32_max.hpp
- device_reduce_instance_threadwise_f32_f32_f32_min.hpp
- device_reduce_instance_threadwise_f32_f32_f32_norm2.hpp
- device_reduce_instance_threadwise_f32_f64_f32_add.hpp
- device_reduce_instance_threadwise_f32_f64_f32_avg.hpp
- device_reduce_instance_threadwise_f32_f64_f32_norm2.hpp
- device_reduce_instance_threadwise_f64_f64_f64_add.hpp
- device_reduce_instance_threadwise_f64_f64_f64_amax.hpp
- device_reduce_instance_threadwise_f64_f64_f64_avg.hpp
- device_reduce_instance_threadwise_f64_f64_f64_max.hpp
- device_reduce_instance_threadwise_f64_f64_f64_min.hpp
- device_reduce_instance_threadwise_f64_f64_f64_norm2.hpp
- device_reduce_instance_threadwise_i8_i32_i8_add.hpp
- device_reduce_instance_threadwise_i8_i32_i8_avg.hpp
- device_reduce_instance_threadwise_i8_i8_i8_amax.hpp
- device_reduce_instance_threadwise_i8_i8_i8_max.hpp
- device_reduce_instance_threadwise_i8_i8_i8_min.hpp
- reduce.hpp
- softmax/
- device_softmax_f16_f16_instance_rank3_reduce1.hpp
- device_softmax_f16_f16_instance_rank3_reduce2.hpp
- device_softmax_f16_f16_instance_rank3_reduce3.hpp
- device_softmax_f16_f16_instance_rank4_reduce1.hpp
- device_softmax_f16_f16_instance_rank4_reduce2.hpp
- device_softmax_f16_f16_instance_rank4_reduce3.hpp
- device_softmax_f16_f16_instance_rank4_reduce4.hpp
- device_softmax_f16_f16_instance_type.hpp
- device_softmax_f32_f32_instance_rank3_reduce1.hpp
- device_softmax_f32_f32_instance_rank3_reduce2.hpp
- device_softmax_f32_f32_instance_rank3_reduce3.hpp
- device_softmax_f32_f32_instance_rank4_reduce1.hpp
- device_softmax_f32_f32_instance_rank4_reduce2.hpp
- device_softmax_f32_f32_instance_rank4_reduce3.hpp
- device_softmax_f32_f32_instance_rank4_reduce4.hpp
- device_softmax_f32_f32_instance_type.hpp
- device_softmax_instance.hpp
- softmax.hpp
- transpose/
- device_transpose_instance.hpp
- transpose_3d.hpp
- utility/
- README.md
- algorithm.hpp
- check_err.hpp
- conv_common.hpp
- convolution_host_tensor_descriptor_helper.hpp
- convolution_parameter.hpp
- device_memory.hpp
- device_tensor_generator.hpp
- fill.hpp
- gpu_verification.hpp
- host_common_util.hpp
- host_gemm.hpp
- host_tensor.hpp
- host_tensor_generator.hpp
- iterator.hpp
- literals.hpp
- numeric.hpp
- ranges.hpp
- thread.hpp
- reference_tensor_operation/
- problem_transform/
- transform_forward_convolution3d_into_gemm_v4r4r4_ndhwc_kzyxc_ndhwk.hpp
- stream_config.hpp
- tensor/
- static_tensor.hpp
- tensor_description/
- cluster_descriptor.hpp
- multi_index_transform.hpp
- multi_index_transform_helper.hpp
- tensor_adaptor.hpp
- tensor_descriptor.hpp
- tensor_descriptor_helper.hpp
- tensor_space_filling_curve.hpp
- tensor_operation/
- gpu/
- block/
- blockwise_gemm_dl_v2r3.hpp
- blockwise_gemm_dlops_v2r2.hpp
- blockwise_gemm_dlops_v3.hpp
- blockwise_gemm_dpp.hpp
- blockwise_gemm_mx_pipeline_xdlops_base.hpp
- blockwise_gemm_pipeline_wmma_selector.hpp
- blockwise_gemm_pipeline_wmmaops.hpp
- blockwise_gemm_pipeline_wmmaops_base.hpp
- blockwise_gemm_pipeline_wmmaops_v1.hpp
- blockwise_gemm_pipeline_wmmaops_v3.hpp
- blockwise_gemm_pipeline_xdlops.hpp
- blockwise_gemm_pipeline_xdlops_ab_scale_selector.hpp
- blockwise_gemm_pipeline_xdlops_b_preshuffle_dequant_v1.hpp
- blockwise_gemm_pipeline_xdlops_b_preshuffle_dequant_v3.hpp
- blockwise_gemm_pipeline_xdlops_b_preshuffle_gufusion_dequant_v1.hpp
- blockwise_gemm_pipeline_xdlops_b_preshuffle_gufusion_v1.hpp
- blockwise_gemm_pipeline_xdlops_b_preshuffle_gufusion_v3.hpp
- blockwise_gemm_pipeline_xdlops_b_preshuffle_mx_moe_gufusion_v3.hpp
- blockwise_gemm_pipeline_xdlops_b_preshuffle_mx_moe_selector.hpp
- blockwise_gemm_pipeline_xdlops_b_preshuffle_mx_moe_v1.hpp
- blockwise_gemm_pipeline_xdlops_b_preshuffle_mx_moe_v3.hpp
- blockwise_gemm_pipeline_xdlops_b_preshuffle_selector.hpp
- blockwise_gemm_pipeline_xdlops_b_preshuffle_v1.hpp
- blockwise_gemm_pipeline_xdlops_b_preshuffle_v2.hpp
- blockwise_gemm_pipeline_xdlops_b_preshuffle_v3.hpp
- blockwise_gemm_pipeline_xdlops_b_scale_selector.hpp
- blockwise_gemm_pipeline_xdlops_base.hpp
- blockwise_gemm_pipeline_xdlops_blockscale_b_preshuffle_selector.hpp
- blockwise_gemm_pipeline_xdlops_blockscale_b_preshuffle_v1.hpp
- blockwise_gemm_pipeline_xdlops_blockscale_b_preshuffle_v3.hpp
- blockwise_gemm_pipeline_xdlops_moe_blockscale_b_preshuffle_gufusion_v1.hpp
- blockwise_gemm_pipeline_xdlops_moe_blockscale_b_preshuffle_gufusion_v3.hpp
- blockwise_gemm_pipeline_xdlops_moe_blockscale_b_preshuffle_selector.hpp
- blockwise_gemm_pipeline_xdlops_moe_blockscale_b_preshuffle_v1.hpp
- blockwise_gemm_pipeline_xdlops_moe_blockscale_b_preshuffle_v3.hpp
- blockwise_gemm_pipeline_xdlops_mx_bpreshuffle_selector.hpp
- blockwise_gemm_pipeline_xdlops_mx_moe_gufusion_v3.hpp
- blockwise_gemm_pipeline_xdlops_mx_moe_nbs_gufusion_v3.hpp
- blockwise_gemm_pipeline_xdlops_mx_moe_nbs_selector.hpp
- blockwise_gemm_pipeline_xdlops_mx_moe_nbs_v1.hpp
- blockwise_gemm_pipeline_xdlops_mx_moe_nbs_v3.hpp
- blockwise_gemm_pipeline_xdlops_mx_moe_selector.hpp
- blockwise_gemm_pipeline_xdlops_mx_moe_v3.hpp
- blockwise_gemm_pipeline_xdlops_mx_selector.hpp
- blockwise_gemm_pipeline_xdlops_selector.hpp
- blockwise_gemm_pipeline_xdlops_v1.hpp
- blockwise_gemm_pipeline_xdlops_v1_ab_scale.hpp
- blockwise_gemm_pipeline_xdlops_v1_b_scale.hpp
- blockwise_gemm_pipeline_xdlops_v1_mx.hpp
- blockwise_gemm_pipeline_xdlops_v2.hpp
- blockwise_gemm_pipeline_xdlops_v2_ab_scale.hpp
- blockwise_gemm_pipeline_xdlops_v2_b_scale.hpp
- blockwise_gemm_pipeline_xdlops_v3.hpp
- blockwise_gemm_pipeline_xdlops_v3_ab_scale.hpp
- blockwise_gemm_pipeline_xdlops_v3_b_scale.hpp
- blockwise_gemm_pipeline_xdlops_v3_mx.hpp
- blockwise_gemm_pipeline_xdlops_v3_mx_bpreshuffle.hpp
- blockwise_gemm_pipeline_xdlops_v4.hpp
- blockwise_gemm_pipeline_xdlops_v4_b_scale.hpp
- blockwise_gemm_pipeline_xdlops_v5.hpp
- blockwise_gemm_smfmac_xdlops.hpp
- blockwise_gemm_wmma.hpp
- blockwise_gemm_xdlops.hpp
- blockwise_gemm_xdlops_skip_b_lds.hpp
- blockwise_softmax.hpp
- blockwise_tensor_slice_transfer_v5r1.hpp
- blockwise_welford.hpp
- reduction_functions_blockwise.hpp
- thread_group_tensor_slice_transfer_direct_load.hpp
- thread_group_tensor_slice_transfer_gather_direct_load.hpp
- thread_group_tensor_slice_transfer_global.hpp
- thread_group_tensor_slice_transfer_v4r1.hpp
- thread_group_tensor_slice_transfer_v4r1_dequant.hpp
- thread_group_tensor_slice_transfer_v4r1_gather.hpp
- thread_group_tensor_slice_transfer_v4r2.hpp
- thread_group_tensor_slice_transfer_v6r1.hpp
- thread_group_tensor_slice_transfer_v6r1r2.hpp
- thread_group_tensor_slice_transfer_v6r2.hpp
- thread_group_tensor_slice_transfer_v6r3.hpp
- thread_group_tensor_slice_transfer_v7.hpp
- thread_group_tensor_slice_transfer_v7r2.hpp
- thread_group_tensor_slice_transfer_v7r3.hpp
- thread_group_tensor_slice_transfer_v7r3_scatter.hpp
- device/
- conv_tensor_rearrange_op.hpp
- convolution_backward_data_specialization.hpp
- convolution_backward_weight_specialization.hpp
- convolution_forward_specialization.hpp
- device_avgpool_bwd.hpp
- device_base.hpp
- device_batched_contraction_multiple_d.hpp
- device_batched_gemm.hpp
- device_batched_gemm_e_permute.hpp
- device_batched_gemm_gemm.hpp
- device_batched_gemm_multi_d.hpp
- device_batched_gemm_multiple_d_gemm_multiple_d.hpp
- device_batched_gemm_softmax_gemm.hpp
- device_batched_gemm_softmax_gemm_permute.hpp
- device_batchnorm_backward.hpp
- device_batchnorm_forward.hpp
- device_batchnorm_infer.hpp
- device_cgemm.hpp
- device_contraction_multiple_abd.hpp
- device_contraction_multiple_d.hpp
- device_conv_bwd_data.hpp
- device_conv_fwd.hpp
- device_conv_fwd_bias_activation.hpp
- device_conv_fwd_bias_activation_add.hpp
- device_conv_tensor_rearrange.hpp
- device_elementwise.hpp
- device_elementwise_normalization.hpp
- device_elementwise_scale.hpp
- device_gemm.hpp
- device_gemm_bias_e_permute.hpp
- device_gemm_dequantB.hpp
- device_gemm_multiple_abd.hpp
- device_gemm_multiple_d.hpp
- device_gemm_multiple_d_ab_scale.hpp
- device_gemm_multiple_d_layernorm.hpp
- device_gemm_multiple_d_multiple_r.hpp
- device_gemm_mx.hpp
- device_gemm_reduce.hpp
- device_gemm_splitk.hpp
- device_gemm_v2.hpp
- device_grouped_contraction_multiple_d.hpp
- device_grouped_conv_bwd_data_multiple_d.hpp
- device_grouped_conv_bwd_weight.hpp
- device_grouped_conv_bwd_weight_multiple_d.hpp
- device_grouped_conv_fwd.hpp
- device_grouped_conv_fwd_multiple_abd.hpp
- device_grouped_conv_fwd_multiple_d.hpp
- device_grouped_gemm.hpp
- device_grouped_gemm_fixed_nk.hpp
- device_grouped_gemm_multi_abd.hpp
- device_grouped_gemm_multi_abd_fixed_nk.hpp
- device_grouped_gemm_softmax_gemm_permute.hpp
- device_grouped_gemm_splitk.hpp
- device_grouped_gemm_tile_loop.hpp
- device_max_pool_bwd.hpp
- device_multiple_reduce.hpp
- device_normalization_bwd_data.hpp
- device_normalization_bwd_gamma_beta.hpp
- device_normalization_fwd.hpp
- device_permute.hpp
- device_pool_fwd.hpp
- device_put_element.hpp
- device_reduce.hpp
- device_reduce_multi_d.hpp
- device_softmax.hpp
- device_splitk_contraction_multiple_d.hpp
- gemm_specialization.hpp
- helper.hpp
- impl/
- codegen_device_grouped_conv_fwd_multiple_abd_xdl_cshuffle.hpp
- device_avgpool2d_bwd_nhwc_nhwc.hpp
- device_avgpool3d_bwd_ndhwc_ndhwc.hpp
- device_batched_contraction_multiple_d_wmma_cshuffle.hpp
- device_batched_contraction_multiple_d_wmma_cshuffle_v3.hpp
- device_batched_contraction_multiple_d_xdl_cshuffle.hpp
- device_batched_gemm_e_permute_xdl.hpp
- device_batched_gemm_gemm_wmma_cshuffle_v3.hpp
- device_batched_gemm_gemm_wmma_cshuffle_v3_common.hpp
- device_batched_gemm_gemm_xdl_cshuffle.hpp
- device_batched_gemm_multi_d_xdl.hpp
- device_batched_gemm_multiple_d_dl.hpp
- device_batched_gemm_multiple_d_gemm_multiple_d_wmma_cshuffle_v3.hpp
- device_batched_gemm_multiple_d_gemm_multiple_d_xdl_cshuffle.hpp
- device_batched_gemm_multiple_d_wmma_cshuffle_v3.hpp
- device_batched_gemm_multiple_d_xdl_cshuffle_v3.hpp
- device_batched_gemm_reduce_wmma_cshuffle_v3.hpp
- device_batched_gemm_reduce_xdl_cshuffle.hpp
- device_batched_gemm_softmax_gemm_permute_wmma_cshuffle.hpp
- device_batched_gemm_softmax_gemm_permute_xdl_cshuffle.hpp
- device_batched_gemm_softmax_gemm_xdl_cshuffle.hpp
- device_batched_gemm_wmma_cshuffle_v3.hpp
- device_batched_gemm_wmma_cshuffle_v3_b_scale.hpp
- device_batched_gemm_wmma_cshuffle_v3_common.hpp
- device_batched_gemm_xdl.hpp
- device_batched_gemm_xdl_fpAintB_b_scale.hpp
- device_batchnorm_backward_impl.hpp
- device_batchnorm_forward_impl.hpp
- device_batchnorm_forward_impl_obsolete.hpp
- device_cgemm_4gemm_xdl_cshuffle.hpp
- device_column_to_image_impl.hpp
- device_contraction_multiple_abd_xdl_cshuffle.hpp
- device_contraction_multiple_d_xdl_cshuffle.hpp
- device_contraction_utils.hpp
- device_conv2d_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp
- device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk.hpp
- device_conv2d_fwd_xdl_c_shuffle_bias_activation_add_nhwc_kyxc_nhwk.hpp
- device_conv2d_fwd_xdl_c_shuffle_bias_activation_nhwc_kyxc_nhwk.hpp
- device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp
- device_conv2d_fwd_xdl_nhwc_kyxc_nhwk.hpp
- device_conv3d_fwd_naive_ndhwc_kzyxc_ndhwk.hpp
- device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk.hpp
- device_convnd_bwd_data_nwc_kxc_nwk_dl.hpp
- device_convnd_bwd_data_nwc_kxc_nwk_xdl.hpp
- device_elementwise_dynamic_vector_dims_impl.hpp
- device_elementwise_normalization_impl.hpp
- device_elementwise_scale_impl.hpp
- device_fpAintB_gemm_wmma.hpp
- device_gemm_bias_add_reduce_wmma_cshuffle_v3.hpp
- device_gemm_bias_add_reduce_xdl_cshuffle.hpp
- device_gemm_dl.hpp
- device_gemm_dpp.hpp
- device_gemm_multiple_abd_wmma_cshuffle_v3.hpp
- device_gemm_multiple_abd_xdl_cshuffle.hpp
- device_gemm_multiple_d_dl.hpp
- device_gemm_multiple_d_layernorm_wmma_cshuffle_v3.hpp
- device_gemm_multiple_d_layernorm_xdl_cshuffle.hpp
- device_gemm_multiple_d_multiple_r_xdl_cshuffle.hpp
- device_gemm_multiple_d_wmma_cshuffle.hpp
- device_gemm_multiple_d_wmma_cshuffle_v3.hpp
- device_gemm_multiple_d_wmma_cshuffle_v3_ab_scale.hpp
- device_gemm_multiple_d_wmma_cshuffle_v3_b_preshuffle.hpp
- device_gemm_multiple_d_wmma_cshuffle_v3_blockscale_bpreshuffle.hpp
- device_gemm_multiple_d_xdl_cshuffle.hpp
- device_gemm_multiple_d_xdl_cshuffle_lds_direct_load.hpp
- device_gemm_multiple_d_xdl_cshuffle_v3.hpp
- device_gemm_multiple_d_xdl_cshuffle_v3_ab_scale.hpp
- device_gemm_multiple_d_xdl_cshuffle_v3_b_preshuffle.hpp
- device_gemm_multiple_d_xdl_cshuffle_v3_blockscale_bpreshuffle.hpp
- device_gemm_reduce_wmma_cshuffle_v3.hpp
- device_gemm_reduce_xdl_cshuffle.hpp
- device_gemm_wmma.hpp
- device_gemm_wmma_cshuffle_v3.hpp
- device_gemm_wmma_cshuffle_v3_b_preshuffle.hpp
- device_gemm_wmma_cshuffle_v3_b_scale.hpp
- device_gemm_wmma_cshuffle_v3_common.hpp
- device_gemm_wmma_cshuffle_v3r1.hpp
- device_gemm_xdl.hpp
- device_gemm_xdl_cshuffle.hpp
- device_gemm_xdl_cshuffle_lds_direct_load.hpp
- device_gemm_xdl_cshuffle_v2.hpp
- device_gemm_xdl_cshuffle_v3.hpp
- device_gemm_xdl_cshuffle_v3_b_preshuffle.hpp
- device_gemm_xdl_cshuffle_v3_b_scale.hpp
- device_gemm_xdl_cshuffle_v3_mx.hpp
- device_gemm_xdl_cshuffle_v3r1.hpp
- device_gemm_xdl_layernorm_cshuffle.hpp
- device_gemm_xdl_skip_b_lds.hpp
- device_gemm_xdl_splitk_c_shuffle.hpp
- device_gemm_xdl_splitk_c_shuffle_lds_direct_load.hpp
- device_gemm_xdl_waveletmodel_cshuffle.hpp
- device_grouped_contraction_multiple_d_xdl_cshuffle.hpp
- device_grouped_conv_bwd_data_multiple_d_wmma_cshuffle.hpp
- device_grouped_conv_bwd_data_multiple_d_wmma_cshuffle_v3.hpp
- device_grouped_conv_bwd_data_multiple_d_xdl_cshuffle_v1.hpp
- device_grouped_conv_bwd_data_multiple_d_xdl_cshuffle_v3.hpp
- device_grouped_conv_bwd_weight_dl.hpp
- device_grouped_conv_bwd_weight_explicit.hpp
- device_grouped_conv_bwd_weight_multiple_d_wmma_cshuffle_v3.hpp
- device_grouped_conv_bwd_weight_multiple_d_xdl_cshuffle.hpp
- device_grouped_conv_bwd_weight_two_stage_wmma_cshuffle_v3.hpp
- device_grouped_conv_bwd_weight_two_stage_xdl_cshuffle.hpp
- device_grouped_conv_bwd_weight_wmma_cshuffle.hpp
- device_grouped_conv_bwd_weight_wmma_cshuffle_v3.hpp
- device_grouped_conv_bwd_weight_xdl_cshuffle.hpp
- device_grouped_conv_bwd_weight_xdl_cshuffle_v3.hpp
- device_grouped_conv_bwd_weight_xdl_waveletmodel_cshuffle_v3.hpp
- device_grouped_conv_fwd_dl_multiple_d_nhwc_kyxc_nhwk.hpp
- device_grouped_conv_fwd_dl_nhwc_kyxc_nhwk.hpp
- device_grouped_conv_fwd_multiple_abd_wmma_cshuffle_v3.hpp
- device_grouped_conv_fwd_multiple_abd_xdl_cshuffle.hpp
- device_grouped_conv_fwd_multiple_abd_xdl_cshuffle_v3.hpp
- device_grouped_conv_fwd_multiple_d_multiple_r.hpp
- device_grouped_conv_fwd_multiple_d_multiple_r_xdl_cshuffle.hpp
- device_grouped_conv_fwd_multiple_d_wmma_cshuffle.hpp
- device_grouped_conv_fwd_multiple_d_wmma_cshuffle_v3_large_tensor.hpp
- device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp
- device_grouped_conv_fwd_multiple_d_xdl_large_tensor_cshuffle.hpp
- device_grouped_conv_utils.hpp
- device_grouped_gemm_fixed_nk_common.hpp
- device_grouped_gemm_multi_abd_wmma_fixed_nk.hpp
- device_grouped_gemm_multi_abd_xdl_fixed_nk.hpp
- device_grouped_gemm_multiple_d_dl.hpp
- device_grouped_gemm_multiple_d_splitk_xdl_cshuffle_two_stage.hpp
- device_grouped_gemm_multiple_d_wmma_cshuffle_tile_loop_v3.hpp
- device_grouped_gemm_multiple_d_xdl_cshuffle_tile_loop.hpp
- device_grouped_gemm_softmax_gemm_permute_xdl_cshuffle.hpp
- device_grouped_gemm_wmma_fixed_nk.hpp
- device_grouped_gemm_wmma_splitk_cshuffle_v3.hpp
- device_grouped_gemm_xdl.hpp
- device_grouped_gemm_xdl_fixed_nk.hpp
- device_grouped_gemm_xdl_splitk_cshuffle.hpp
- device_grouped_query_attention_forward_wmma.hpp
- device_image_to_column_impl.hpp
- device_max_pool_bwd_impl.hpp
- device_moe_gemm.hpp
- device_moe_gemm_blockscale.hpp
- device_moe_mx_gemm.hpp
- device_moe_mx_gemm_bns.hpp
- device_moe_mx_gemm_bpreshuffle.hpp
- device_multi_query_attention_forward_wmma.hpp
- device_multiple_reduce_multiblock.hpp
- device_multiple_reduce_threadwise.hpp
- device_normalization_bwd_data_impl.hpp
- device_normalization_bwd_gamma_beta_impl.hpp
- device_normalization_fwd_impl.hpp
- device_normalization_fwd_splitk_impl.hpp
- device_permute_impl.hpp
- device_pool2d_fwd_nhwc_nhwc.hpp
- device_pool3d_fwd_ndhwc_ndhwc.hpp
- device_put_element_impl.hpp
- device_reduce_common.hpp
- device_reduce_multiblock.hpp
- device_reduce_threadwise.hpp
- device_reduce_threadwise_multi_d.hpp
- device_softmax_impl.hpp
- device_sparse_embeddings_forward_layernorm.hpp
- device_splitk_contraction_multiple_d_xdl_cshuffle.hpp
- split_k_arg.hpp
- split_k_offset_utils.hpp
- split_k_utils.hpp
- masking_specialization.hpp
- matrix_padder.hpp
- reduction_operator_mapping.hpp
- tensor_layout.hpp
- tensor_size_check.hpp
- tensor_specialization.hpp
- welford_helper.hpp
- element/
- binary_element_wise_operation.hpp
- combined_element_wise_operation.hpp
- element_wise_operation.hpp
- quantization_operation.hpp
- unary_element_wise_operation.hpp
- grid/
- batchnorm_multiblock/
- gridwise_multiblock_batchnorm_forward.hpp
- gridwise_multiblock_reduce_second_half_batchnorm_backward_final.hpp
- gridwise_multiblock_welford_first_half.hpp
- gridwise_multiblock_welford_second_half_batchnorm_forward_final_obsolete.hpp
- gridwise_multiblock_welford_second_half_multiblock_reduce_first_half.hpp
- block_to_ctile_map.hpp
- epilogue_cshuffle_v3_reduce_wmma.hpp
- epilogue_cshuffle_v3_welford_wmma.hpp
- epilogue_cshuffle_v3_wmma.hpp
- epilogue_cshuffle_v3_wmma_base.hpp
- epilogue_direct_store.hpp
- epilogue_type.hpp
- gemm_layernorm/
- gridwise_gemm_multiple_d_welford_first_half_xdl_cshuffle.hpp
- gridwise_welford_second_half_layernorm2d.hpp
- gridwise_2d_multiple_reduction_multiblock.hpp
- gridwise_2d_multiple_reduction_threadwise.hpp
- gridwise_2d_reduction_multiblock.hpp
- gridwise_2d_reduction_threadwise.hpp
- gridwise_2d_reduction_threadwise_multi_d.hpp
- gridwise_ab_transfer_thread_tiles.hpp
- gridwise_ab_transfer_thread_tiles_preshuffle.hpp
- gridwise_ab_transfer_wave_tiles.hpp
- gridwise_ab_transfer_wave_tiles_interleave.hpp
- gridwise_batched_gemm_gemm_wmma_cshuffle_v3.hpp
- gridwise_batched_gemm_gemm_xdl_cshuffle_v1.hpp
- gridwise_batched_gemm_multiple_d_gemm_multiple_d_xdl_cshuffle_v1.hpp
- gridwise_batched_gemm_multiple_d_softmax_gemm_xdl_cshuffle_v1.hpp
- gridwise_batched_gemm_softmax_gemm_wmma_cshuffle.hpp
- gridwise_batched_gemm_softmax_gemm_xdl_cshuffle_v1.hpp
- gridwise_batchnorm_backward_blockwise_welford.hpp
- gridwise_batchnorm_forward_blockwise_welford.hpp
- gridwise_elementwise_1d_scale.hpp
- gridwise_elementwise_2d.hpp
- gridwise_elementwise_layernorm_welford_variance.hpp
- gridwise_fpAintB_gemm_wmma.hpp
- gridwise_gemm_bias_add_reduce_xdl_cshuffle_v1.hpp
- gridwise_gemm_dl_multiple_d.hpp
- gridwise_gemm_dl_v1r3.hpp
- gridwise_gemm_dpp.hpp
- gridwise_gemm_multiple_abd_xdl_cshuffle.hpp
- gridwise_gemm_multiple_d_multiple_r_xdl_cshuffle.hpp
- gridwise_gemm_multiple_d_wmma_cshuffle.hpp
- gridwise_gemm_multiple_d_xdl_cshuffle.hpp
- gridwise_gemm_multiple_d_xdl_cshuffle_lds_direct_load.hpp
- gridwise_gemm_multiple_d_xdl_splitk_cshuffle.hpp
- 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
- gridwise_gemm_split_k_multiple_d_xdl_cshuffle.hpp
- gridwise_gemm_split_k_multiple_d_xdl_cshuffle_v2.hpp
- gridwise_gemm_waveletmodel.hpp
- gridwise_gemm_wmma.hpp
- gridwise_gemm_wmma_cshuffle_v3.hpp
- gridwise_gemm_wmma_cshuffle_v3_ab_scale.hpp
- gridwise_gemm_wmma_cshuffle_v3_common.hpp
- gridwise_gemm_wmma_cshuffle_v3_common_kernels.hpp
- gridwise_gemm_xdl_cshuffle_common.hpp
- gridwise_gemm_xdl_cshuffle_conv_v3.hpp
- gridwise_gemm_xdl_cshuffle_v1.hpp
- gridwise_gemm_xdl_cshuffle_v2.hpp
- gridwise_gemm_xdl_cshuffle_v3.hpp
- gridwise_gemm_xdl_cshuffle_v3_b_preshuffle.hpp
- gridwise_gemm_xdl_cshuffle_v3_b_scale.hpp
- gridwise_gemm_xdl_cshuffle_v3_multi_abd.hpp
- gridwise_gemm_xdl_cshuffle_v3_multi_d.hpp
- gridwise_gemm_xdl_cshuffle_v3_multi_d_ab_scale.hpp
- gridwise_gemm_xdl_cshuffle_v3_multi_d_b_preshuffle.hpp
- gridwise_gemm_xdl_cshuffle_v3_multi_d_blockscale_b_preshuffle.hpp
- gridwise_gemm_xdl_cshuffle_v3_mx.hpp
- gridwise_gemm_xdl_cshuffle_v3_mx_bpreshuffle.hpp
- gridwise_gemm_xdl_layernorm_cshuffle_v1.hpp
- gridwise_gemm_xdl_waveletmodel_cshuffle.hpp
- gridwise_gemm_xdl_waveletmodel_cshuffle_conv_v3.hpp
- gridwise_gemm_xdlops_bwd_weight.hpp
- gridwise_gemm_xdlops_skip_b_lds_v1.hpp
- gridwise_gemm_xdlops_splitk_lds_direct_load.hpp
- gridwise_gemm_xdlops_v2r3.hpp
- gridwise_gemm_xdlops_v2r4.hpp
- gridwise_gemm_xdlops_v2r4r2.hpp
- gridwise_gemm_xdlops_v3r1.hpp
- gridwise_gemm_xdlops_v3r2.hpp
- gridwise_gemm_xdlops_v3r3.hpp
- gridwise_moe_gemm.hpp
- gridwise_moe_gemm_blockscale.hpp
- gridwise_moe_mx_gemm.hpp
- gridwise_moe_mx_gemm_bns.hpp
- gridwise_moe_mx_gemm_bpreshuffle.hpp
- 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.hpp
- gridwise_sparse_embeddings_forward_layernorm_builtins.hpp
- gridwise_tensor_rearrange.hpp
- normalization/
- gridwise_normalization_bwd_data.hpp
- gridwise_normalization_bwd_gamma_beta.hpp
- gridwise_normalization_naive_variance.hpp
- gridwise_normalization_selector.hpp
- gridwise_normalization_splitk_1st.hpp
- gridwise_normalization_splitk_2nd.hpp
- gridwise_normalization_welford_variance.hpp
- batchnorm_multiblock/
- thread/
- reduction_functions_threadwise.hpp
- threadwise_contraction_dl.hpp
- threadwise_gemm_dlops_v3.hpp
- threadwise_tensor_slice_set.hpp
- threadwise_tensor_slice_transfer.hpp
- threadwise_tensor_slice_transfer_util.hpp
- threadwise_tensor_slice_transfer_v3r1.hpp
- threadwise_tensor_slice_transfer_v3r1_dequant.hpp
- threadwise_tensor_slice_transfer_v3r1_gather.hpp
- threadwise_tensor_slice_transfer_v3r2.hpp
- threadwise_tensor_slice_transfer_v4r1.hpp
- threadwise_tensor_slice_transfer_v5r1.hpp
- threadwise_tensor_slice_transfer_v6r1.hpp
- threadwise_tensor_slice_transfer_v6r1r2.hpp
- threadwise_tensor_slice_transfer_v6r2.hpp
- threadwise_tensor_slice_transfer_v6r3.hpp
- threadwise_tensor_slice_transfer_v7.hpp
- threadwise_tensor_slice_transfer_v7r2.hpp
- threadwise_tensor_slice_transfer_v7r3.hpp
- threadwise_tensor_slice_transfer_v7r3_scatter.hpp
- threadwise_welford.hpp
- warp/
- dpp_gemm.hpp
- smfmac_xdlops_gemm.hpp
- wmma_gemm.hpp
- xdlops_gemm.hpp
- block/
- operator_transform/
- transform_contraction_to_gemm.hpp
- transform_contraction_to_gemm_arraybase.hpp
- transform_conv_bwd_data_to_gemm_v1.hpp
- transform_conv_bwd_weight_to_gemm.hpp
- transform_conv_bwd_weight_to_gemm_v2.hpp
- transform_conv_fwd_to_gemm.hpp
- transform_conv_ngchw_to_nhwgc.hpp
- gpu/
- utility/
- amd_address_space.hpp
- amd_arch.hpp
- amd_buffer_addressing.hpp
- amd_buffer_addressing_builtins.hpp
- amd_buffer_coherence.hpp
- amd_ck_fp8.hpp
- amd_cluster_load.hpp
- amd_gemm_dpp.hpp
- amd_inline_asm.hpp
- amd_lds.hpp
- amd_smfmac.hpp
- amd_transpose_load.hpp
- amd_wave_read_first_lane.hpp
- amd_wmma.hpp
- amd_xdlops.hpp
- array.hpp
- array_multi_index.hpp
- blkgemmpipe_scheduler.hpp
- c_style_pointer_cast.hpp
- common_header.hpp
- container_element_picker.hpp
- container_helper.hpp
- data_cache_prefetch.hpp
- data_type.hpp
- debug.hpp
- dtype_fp64.hpp
- dtype_vector.hpp
- dynamic_buffer.hpp
- e4m3.hpp
- e5m3.hpp
- e8m0.hpp
- enable_if.hpp
- env.hpp
- f8_utils.hpp
- filter_tuple.hpp
- flush_icache.hpp
- functional.hpp
- functional2.hpp
- functional3.hpp
- functional4.hpp
- generic_memory_space_atomic.hpp
- get_id.hpp
- get_shift.hpp
- ignore.hpp
- index_expression.hpp
- inner_product.hpp
- inner_product_dpp8.hpp
- integral_constant.hpp
- is_detected.hpp
- is_known_at_compile_time.hpp
- loop_scheduler.hpp
- magic_division.hpp
- math.hpp
- math_v2.hpp
- multi_index.hpp
- mxf4_utils.hpp
- mxf6_utils.hpp
- mxf8_utils.hpp
- mxfp_utils.hpp
- number.hpp
- numeric_limits.hpp
- numeric_utils.hpp
- pipeline_enum.hpp
- random_gen.hpp
- reduction_common.hpp
- reduction_enums.hpp
- reduction_functions_accumulate.hpp
- reduction_operator.hpp
- scale_utils.hpp
- scaled_type_convert.hpp
- scheduler_enum.hpp
- sequence.hpp
- sequence_helper.hpp
- span.hpp
- static_buffer.hpp
- statically_indexed_array.hpp
- statically_indexed_array_multi_index.hpp
- synchronization.hpp
- thread_buf_to_vec_loader.hpp
- thread_group.hpp
- transpose_vectors.hpp
- tuple.hpp
- tuple_helper.hpp
- type.hpp
- type_convert.hpp
- workgroup_barrier.hpp
- workgroup_synchronization.hpp
- version.h
- version.h.in
- wrapper/
- layout.hpp
- operations/
- copy.hpp
- gemm.hpp
- tensor.hpp
- traits/
- blockwise_gemm_xdl_traits.hpp
- utils/
- kernel_utils.hpp
- layout_utils.hpp
- tensor_partition.hpp
- tensor_utils.hpp
- ck_tile/
- README.md
- core/
- README.md
- algorithm/
- cluster_descriptor.hpp
- coordinate_transform.hpp
- indexing_adaptor.hpp
- space_filling_curve.hpp
- static_encoding_pattern.hpp
- arch/
- amd_buffer_addressing.hpp
- amd_buffer_addressing_builtins.hpp
- amd_buffer_coherence.hpp
- amd_cluster_load.hpp
- amd_tdm_descriptor.hpp
- amd_transpose_load_encoding.hpp
- amd_wave_read_first_lane.hpp
- arch.hpp
- barrier.hpp
- generic_memory_space_atomic.hpp
- inst_prefetch.hpp
- mma/
- amdgcn_mma.hpp
- mfma/
- mfma.hpp
- mfma_gfx9.hpp
- mfma_selector.hpp
- mfma_traits.hpp
- mfma_transforms.hpp
- mma.hpp
- mma_op_family.hpp
- mma_pipeline.hpp
- mma_selector.hpp
- mma_traits.hpp
- mma_transforms.hpp
- mma_wavewise.hpp
- scale/
- mfma/
- scale_gfx9.hpp
- selector.hpp
- scale.hpp
- scale_mma_pipeline.hpp
- scale_selector.hpp
- scale_traits.hpp
- scale_transforms.hpp
- mfma/
- sparse/
- mfma/
- selector.hpp
- sparse_gfx9.hpp
- sparse.hpp
- sparse_mma_pipeline.hpp
- sparse_selector.hpp
- sparse_traits.hpp
- sparse_transforms.hpp
- wmma/
- selector.hpp
- sparse_gfx12.hpp
- mfma/
- utility/
- tile_distribution_encoding_calculator.hpp
- tile_distribution_encoding_register_mapper.hpp
- wmma/
- wmma.hpp
- wmma_gfx11.hpp
- wmma_gfx12.hpp
- wmma_selector.hpp
- wmma_traits.hpp
- wmma_transforms.hpp
- utility.hpp
- workgroup_barrier.hpp
- config.hpp
- container/
- array.hpp
- container_helper.hpp
- map.hpp
- meta_data_buffer.hpp
- multi_index.hpp
- sequence.hpp
- span.hpp
- static_array.hpp
- statically_indexed_array.hpp
- thread_buffer.hpp
- tuple.hpp
- numeric/
- bfloat16.hpp
- e4m3.hpp
- e5m3.hpp
- e8m0.hpp
- ext_vector_base.hpp
- float8.hpp
- float8_ext.hpp
- half.hpp
- int8.hpp
- integer.hpp
- integral_constant.hpp
- math.hpp
- math_v2.hpp
- mxfp_convert.hpp
- mxfp_scale.hpp
- null_type.hpp
- numeric.hpp
- pk_f6.hpp
- pk_fp4.hpp
- pk_int4.hpp
- scale_util.hpp
- tfloat32.hpp
- type_convert.hpp
- vector_type.hpp
- tensor/
- buffer_view.hpp
- load_tile.hpp
- load_tile_transpose.hpp
- null_tensor.hpp
- null_tile_window.hpp
- shuffle_tile.hpp
- slice_tile.hpp
- static_distributed_tensor.hpp
- store_tile.hpp
- sweep_tile.hpp
- tensor_adaptor.hpp
- tensor_adaptor_coordinate.hpp
- tensor_coordinate.hpp
- tensor_descriptor.hpp
- tensor_view.hpp
- tile_distribution.hpp
- tile_distribution_encoding.hpp
- tile_elementwise.hpp
- tile_scatter_gather.hpp
- tile_window.hpp
- tile_window_base.hpp
- tile_window_linear.hpp
- tile_window_utils.hpp
- transpose_tile.hpp
- update_tile.hpp
- utility/
- bit_cast.hpp
- data_cache_prefetch.hpp
- debug.hpp
- env.hpp
- functional.hpp
- functional_with_tuple.hpp
- gemm_validation.hpp
- ignore.hpp
- literals.hpp
- magic_div.hpp
- mixed_prec_compute_type.hpp
- persistent_async_input_scheduler.hpp
- philox_rand.hpp
- print.hpp
- random.hpp
- reduce_operator.hpp
- reduce_operator_accumulate.hpp
- static_counter.hpp
- to_sequence.hpp
- transpose_vectors.hpp
- type_traits.hpp
- unary_element_function.hpp
- core.hpp
- host/
- arg_parser.hpp
- check_err.hpp
- concat.hpp
- convolution_host_tensor_descriptor_helper.hpp
- convolution_parameter.hpp
- device_memory.hpp
- device_prop.hpp
- fill.hpp
- flush_icache.hpp
- high_res_cpu_clock.hpp
- hip_check_error.hpp
- host_tensor.hpp
- host_tensor_io.hpp
- joinable_thread.hpp
- kernel_launch.hpp
- mx_processing.hpp
- permute_pk_int4.hpp
- pinned_host_releaser.hpp
- ranges.hpp
- reference/
- reference_batched_contraction.hpp
- reference_batched_dropout.hpp
- reference_batched_dropout_randval.hpp
- reference_batched_elementwise.hpp
- reference_batched_gemm.hpp
- reference_batched_masking.hpp
- reference_batched_mx_descale.hpp
- reference_batched_rotary_position_embedding.hpp
- reference_batched_softmax.hpp
- reference_batched_transpose.hpp
- reference_blocked_attention.hpp
- reference_elementwise.hpp
- reference_fused_moe.hpp
- reference_gemm.hpp
- reference_grouped_conv_bwd_data.hpp
- reference_grouped_conv_bwd_weight.hpp
- reference_grouped_conv_fwd.hpp
- reference_im2col.hpp
- reference_layernorm2d_fwd.hpp
- reference_moe_gemm.hpp
- reference_moe_sorting.hpp
- reference_permute.hpp
- reference_pool.hpp
- reference_reduce.hpp
- reference_rmsnorm2d_fwd.hpp
- reference_rowwise_quantization2d.hpp
- reference_softmax.hpp
- reference_topk.hpp
- reference_transpose.hpp
- rotating_buffers.hpp
- stream_config.hpp
- stream_utils.hpp
- tensor_shuffle_utils.hpp
- timer.hpp
- host.hpp
- ops/
- add_rmsnorm2d_rdquant/
- kernel/
- add_rmsnorm2d_rdquant_fwd_kernel.hpp
- pipeline/
- add_rmsnorm2d_rdquant_fwd_pipeline_default_policy.hpp
- add_rmsnorm2d_rdquant_fwd_pipeline_one_pass.hpp
- add_rmsnorm2d_rdquant_fwd_pipeline_problem.hpp
- add_rmsnorm2d_rdquant_fwd_pipeline_three_pass.hpp
- kernel/
- add_rmsnorm2d_rdquant.hpp
- batched_contraction/
- kernel/
- batched_contraction_kernel.hpp
- pipeline/
- batched_contraction_problem.hpp
- utils/
- tensor_descriptor_utils.hpp
- kernel/
- batched_contraction.hpp
- batched_transpose/
- kernel/
- batched_transpose_kernel.hpp
- pipeline/
- batched_transpose_common_policy.hpp
- batched_transpose_lds_pipeline.hpp
- batched_transpose_lds_policy.hpp
- batched_transpose_lds_problem.hpp
- batched_transpose_pipeline.hpp
- batched_transpose_policy.hpp
- batched_transpose_problem.hpp
- kernel/
- batched_transpose.hpp
- bias.hpp
- common/
- README.md
- generic_2d_block_shape.hpp
- load_and_convert_tile.hpp
- streamk_common.hpp
- tensor_layout.hpp
- utils.hpp
- common.hpp
- elementwise/
- binary_elementwise_operation.hpp
- kernel/
- elementwise_kernel.hpp
- pipeline/
- elementwise_pipeline_default_policy.hpp
- elementwise_pipeline_problem.hpp
- elementwise_shape.hpp
- unary_element_wise_operation.hpp
- elementwise.hpp
- epilogue/
- chainer/
- README.md
- common_epilogue_ops.hpp
- cshuffle_epilogue_chainer_ops.hpp
- cshuffle_epilogue_schedule.hpp
- epilogue_chainer.hpp
- cshuffle_epilogue.hpp
- default_2d_and_dynamic_quant_epilogue.hpp
- default_2d_epilogue.hpp
- dynamic_quant_epilogue.hpp
- permuten_epilogue.hpp
- tdm_epilogue.hpp
- chainer/
- epilogue.hpp
- flatmm/
- block/
- block_flatmm_asmem_bsmem_creg_v1.hpp
- block_flatmm_asmem_bsmem_creg_v1_custom_policy.hpp
- flatmm_32x512x128_1x4x1_16x16x32.hpp
- flatmm_sn_32x128x512_1x4x1_16x16x32.hpp
- flatmm_sn_32x128x512_1x4x1_16x16x32_itl.hpp
- flatmm_uk_config.hpp
- uk/
- README.md
- flatmm_sn_uk_gfx9_32x128x512_1x4x1_16x16x16.inc
- flatmm_sn_uk_gfx9_32x128x512_1x4x1_16x16x16_itl.inc
- flatmm_uk_gfx9_32x512x128_1x1x1_16x16x16.inc
- kernel/
- flatmm_kernel.hpp
- grouped_flatmm_kernel.hpp
- grouped_mx_flatmm_kernel.hpp
- mixed_prec_flatmm_kernel.hpp
- moe_flatmm_kernel.hpp
- mx_flatmm_kernel.hpp
- pipeline/
- flatmm_pipeline_agmem_bgmem_creg_v1.hpp
- flatmm_pipeline_agmem_bgmem_creg_v1_policy.hpp
- mixed_prec_flatmm_pipeline_agmem_bgmem_creg_v1.hpp
- mixed_prec_flatmm_pipeline_agmem_bgmem_creg_v1_policy.hpp
- moe_flatmm_pipeline_agmem_bgmem_creg.hpp
- mx_flatmm_pipeline_agmem_bgmem_creg_v1.hpp
- mx_flatmm_pipeline_agmem_bgmem_creg_v1_policy.hpp
- tile_flatmm_shape.hpp
- block/
- flatmm.hpp
- fmha/
- block/
- block_attention_bias_enum.hpp
- block_attention_kv_load_mode_enum.hpp
- block_attention_kvcache_layout_enum.hpp
- block_attention_quant_scale_enum.hpp
- block_dropout.hpp
- block_masking.hpp
- block_position_encoding.hpp
- block_rotary_embedding.hpp
- cast_tile_mx.hpp
- page_block_navigator.hpp
- variants.hpp
- kernel/
- fmha_batch_prefill_kernel.hpp
- fmha_bwd_kernel.hpp
- fmha_fwd_appendkv_kernel.hpp
- fmha_fwd_appendkv_tile_partitioner.hpp
- fmha_fwd_kernel.hpp
- fmha_fwd_pagedkv_kernel.hpp
- fmha_fwd_splitkv_combine_kernel.hpp
- fmha_fwd_splitkv_kernel.hpp
- fmha_fwd_v3_kernel.hpp
- pipeline/
- block_fmha_batch_prefill_pipeline_qr_ks_vs_async.hpp
- block_fmha_batch_prefill_pipeline_qr_ks_vs_async_default_policy.hpp
- block_fmha_bwd_convert_dq.hpp
- block_fmha_bwd_dot_do_o.hpp
- block_fmha_bwd_dq_dk_dv_pipeline_kr_ktr_vr.hpp
- block_fmha_bwd_dq_dk_dv_pipeline_kr_ktr_vr_iglp.hpp
- block_fmha_bwd_dq_dk_dv_pipeline_selector.hpp
- block_fmha_bwd_dq_dk_dv_pipeline_trload_kr_ktr_vr.hpp
- block_fmha_bwd_dq_dk_dv_pipeline_trload_qr_qtr_dor.hpp
- block_fmha_bwd_pipeline_default_policy.hpp
- block_fmha_bwd_pipeline_problem.hpp
- block_fmha_bwd_pipeline_trload_default_policy.hpp
- block_fmha_fwd_appendkv_pipeline.hpp
- block_fmha_fwd_appendkv_pipeline_default_policy.hpp
- block_fmha_fwd_pagedkv_pipeline_qr_ks_vs.hpp
- block_fmha_fwd_pagedkv_pipeline_qr_ks_vs_default_policy.hpp
- block_fmha_fwd_splitkv_combine_pipeline.hpp
- block_fmha_fwd_splitkv_combine_pipeline_default_policy.hpp
- block_fmha_fwd_splitkv_pipeline_nwarp_sshuffle_qr_ks_vs.hpp
- block_fmha_fwd_splitkv_pipeline_nwarp_sshuffle_qr_ks_vs_default_policy.hpp
- block_fmha_fwd_splitkv_pipeline_qr_ks_vs.hpp
- block_fmha_fwd_splitkv_pipeline_qr_ks_vs_default_policy.hpp
- block_fmha_fwd_v3_pipeline.hpp
- block_fmha_fwd_v3_pipeline_default_policy.hpp
- block_fmha_pipeline_enum.hpp
- block_fmha_pipeline_problem.hpp
- block_fmha_pipeline_qr_ks_vs.hpp
- block_fmha_pipeline_qr_ks_vs_async.hpp
- block_fmha_pipeline_qr_ks_vs_async_default_policy.hpp
- block_fmha_pipeline_qr_ks_vs_async_trload.hpp
- block_fmha_pipeline_qr_ks_vs_async_trload_policy.hpp
- block_fmha_pipeline_qr_ks_vs_default_policy.hpp
- block_fmha_pipeline_qr_ks_vs_fp8.hpp
- block_fmha_pipeline_qr_ks_vs_whole_k_prefetch.hpp
- block_fmha_pipeline_qr_ks_vs_whole_k_prefetch_default_policy.hpp
- block_fmha_pipeline_qr_ks_vs_whole_k_prefetch_trload.hpp
- block_fmha_pipeline_qs_ks_vs.hpp
- block_fmha_pipeline_qs_ks_vs_default_policy.hpp
- block_fmha_pipeline_qx_ks_vs_custom_policy.hpp
- tile_fmha_shape.hpp
- tile_fmha_traits.hpp
- block/
- fmha.hpp
- fmha_bwd.hpp
- fmha_bwd_runner.hpp
- fmha_fwd.hpp
- fmha_fwd_head_grouping.hpp
- fmha_fwd_runner.hpp
- fused_moe/
- kernel/
- fused_moegemm_kernel.hpp
- fused_moegemm_shape.hpp
- fused_moegemm_tile_partitioner.hpp
- moe_sorting_kernel.hpp
- moe_sorting_problem.hpp
- pipeline/
- fused_moegemm_pipeline_flatmm_ex.hpp
- fused_moegemm_pipeline_flatmm_policy.hpp
- fused_moegemm_pipeline_flatmm_uk.hpp
- fused_moegemm_pipeline_problem.hpp
- fused_moegemm_traits.hpp
- moe_sorting_pipeline.hpp
- moe_sorting_policy.hpp
- kernel/
- fused_moe.hpp
- gemm/
- block/
- block_gemm_areg_bgmem_creg_v1.hpp
- block_gemm_areg_bgmem_creg_v1_default_policy.hpp
- block_gemm_areg_breg_creg_eight_waves_v1.hpp
- block_gemm_areg_breg_creg_v1.hpp
- block_gemm_areg_breg_creg_v1_custom_policy.hpp
- block_gemm_areg_breg_creg_v1_default_policy.hpp
- block_gemm_areg_breg_creg_v2.hpp
- block_gemm_areg_breg_creg_v2_custom_policy.hpp
- block_gemm_areg_bsmem_creg_one_warp_v1.hpp
- block_gemm_areg_bsmem_creg_v1.hpp
- block_gemm_areg_bsmem_creg_v1_custom_policy.hpp
- block_gemm_areg_bsmem_creg_v1_default_policy.hpp
- block_gemm_areg_bsmem_creg_v2.hpp
- block_gemm_areg_bsmem_creg_v2_custom_policy.hpp
- block_gemm_areg_bsmem_creg_v2_default_policy.hpp
- block_gemm_areg_bsmem_creg_v2_prefetch_k.hpp
- block_gemm_areg_bsmem_creg_v2_prefetch_n.hpp
- block_gemm_areg_bsmem_creg_v2r1.hpp
- block_gemm_areg_bsmem_trload_creg_v2_prefetch_n.hpp
- block_gemm_asmem_breg_creg_v1.hpp
- block_gemm_asmem_breg_creg_v1_custom_policy.hpp
- block_gemm_asmem_breg_creg_v1_default_policy.hpp
- block_gemm_asmem_bsmem_creg_v1.hpp
- block_gemm_asmem_bsmem_creg_v1_custom_policy.hpp
- block_gemm_asmem_bsmem_creg_v1_default_policy.hpp
- block_gemm_mx_areg_bsmem_creg_v1.hpp
- block_gemm_mx_areg_bsmem_creg_v1_custom_policy.hpp
- block_gemm_problem.hpp
- block_universal_gemm_as_bs_cr.hpp
- block_wp_asmem_breg_creg.hpp
- block_wp_asmem_bsmem_creg_v1.hpp
- block_wp_asmem_bsmem_creg_v1_custom_policy.hpp
- kernel/
- batched_gemm_kernel.hpp
- gemm_kernel.hpp
- gemm_multi_abd_kernel.hpp
- gemm_multi_d_kernel.hpp
- gemm_tile_partitioner.hpp
- grouped_gemm_kernel.hpp
- mx_gemm_kernel.hpp
- mx_grouped_gemm_kernel.hpp
- streamk_gemm/
- streamk_gemm_coherency.hpp
- streamk_gemm_kernel.hpp
- streamk_gemm_tile_partitioner.hpp
- streamk_gemm_tile_partitioner_impl.hpp
- universal_gemm_kernel.hpp
- pipeline/
- gemm_pipeline_ag_bg_cr_base.hpp
- gemm_pipeline_ag_bg_cr_comp_async.hpp
- gemm_pipeline_ag_bg_cr_comp_async_default_policy.hpp
- gemm_pipeline_ag_bg_cr_comp_async_eight_waves.hpp
- gemm_pipeline_ag_bg_cr_comp_async_eight_waves_policy.hpp
- gemm_pipeline_ag_bg_cr_comp_async_v2.hpp
- gemm_pipeline_ag_bg_cr_comp_tdm_default_policy.hpp
- gemm_pipeline_ag_bg_cr_comp_tdm_v1.hpp
- gemm_pipeline_ag_bg_cr_comp_tdm_v2.hpp
- gemm_pipeline_ag_bg_cr_comp_v3.hpp
- gemm_pipeline_ag_bg_cr_comp_v4.hpp
- gemm_pipeline_ag_bg_cr_comp_v4_default_policy.hpp
- gemm_pipeline_ag_bg_cr_comp_v5.hpp
- gemm_pipeline_ag_bg_cr_comp_v5_default_policy.hpp
- gemm_pipeline_ag_bg_cr_comp_v6.hpp
- gemm_pipeline_ag_bg_cr_comp_v6_default_policy.hpp
- gemm_pipeline_ag_bg_cr_default_policy.hpp
- gemm_pipeline_ag_bg_cr_eight_waves_base.hpp
- gemm_pipeline_ag_bg_cr_mem.hpp
- gemm_pipeline_ag_bg_cr_scheduler.hpp
- gemm_pipeline_ag_bg_cr_wavelet.hpp
- gemm_pipeline_agmem_bgmem_creg_async_v1.hpp
- gemm_pipeline_agmem_bgmem_creg_v1.hpp
- gemm_pipeline_agmem_bgmem_creg_v1_default_policy.hpp
- gemm_pipeline_agmem_bgmem_creg_v2.hpp
- gemm_pipeline_agmem_bgmem_creg_v2_default_policy.hpp
- gemm_pipeline_problem.hpp
- gemm_pipelines.hpp
- gemm_universal_pipeline_ag_bg_cr_policy.hpp
- tile_gemm_shape.hpp
- tile_gemm_traits.hpp
- wp_pipeline_agmem_bgmem_creg_base_policy.hpp
- wp_pipeline_agmem_bgmem_creg_tdm.hpp
- wp_pipeline_agmem_bgmem_creg_tdm_policy.hpp
- wp_pipeline_agmem_bgmem_creg_v2.hpp
- warp/
- warp_gemm.hpp
- warp_gemm_attribute_mfma.hpp
- warp_gemm_attribute_mfma_impl.hpp
- warp_gemm_attribute_smfmac.hpp
- warp_gemm_attribute_smfmac_impl.hpp
- warp_gemm_attribute_wmma.hpp
- warp_gemm_attribute_wmma_impl.hpp
- warp_gemm_attribute_wmma_impl_16bit_traits.hpp
- warp_gemm_attribute_wmma_impl_8bit_traits.hpp
- warp_gemm_attribute_wmma_impl_base_traits.hpp
- warp_gemm_attribute_wmma_impl_highprec_traits.hpp
- warp_gemm_dispatcher.hpp
- warp_gemm_dispatcher_unification.hpp
- warp_gemm_impl.hpp
- warp_gemm_params.hpp
- warp_gemm_smfmac_impl.hpp
- warp_wmma_gemm.hpp
- warp_wmma_gemm_gfx11_utils.hpp
- block/
- gemm.hpp
- gemm_mx/
- README.md
- block/
- block_mx_asmem_breg_creg.hpp
- block_mx_gemm_areg_breg_creg_eight_waves_v1.hpp
- block_mx_gemm_areg_breg_creg_v1.hpp
- kernel/
- gemm_mx_kernel.hpp
- scale_pointer.hpp
- pipeline/
- gemm_pipeline_ag_bg_cr_comp_async.hpp
- gemm_pipeline_ag_bg_cr_comp_async_default_policy.hpp
- gemm_pipeline_ag_bg_cr_comp_async_eight_waves.hpp
- gemm_pipeline_ag_bg_cr_comp_async_eight_waves_policy.hpp
- wp_pipeline_agmem_bgmem_creg_v1.hpp
- wp_pipeline_agmem_bgmem_creg_v1_policy.hpp
- gemm_mx.hpp
- gemm_quant/
- block/
- block_gemm_quant_common.hpp
- block_universal_gemm_ar_aquant_flatbr_bquant_cr.hpp
- block_universal_gemm_ar_flatbr_bquant_cr.hpp
- block_universal_gemm_as_aquant_bs_bquant_cr.hpp
- block_universal_gemm_as_aquant_bs_bquant_cr_eight_waves.hpp
- block_universal_gemm_as_aquant_bs_cr.hpp
- block_universal_gemm_as_bs_bquant_cr.hpp
- kernel/
- gemm_quant_kernel.hpp
- grouped_gemm_quant_kernel.hpp
- pipeline/
- gemm_abquant_pipeline_ag_bg_cr_base.hpp
- gemm_abquant_pipeline_ag_bg_cr_eight_waves.hpp
- gemm_abquant_pipeline_ag_bg_cr_eight_waves_policy.hpp
- gemm_abquant_pipeline_ag_bg_cr_policy.hpp
- gemm_abquant_pipeline_ag_bg_cr_v3.hpp
- gemm_aquant_pipeline_ag_bg_cr_base.hpp
- gemm_aquant_pipeline_ag_bg_cr_mem.hpp
- gemm_aquant_pipeline_ag_bg_cr_policy.hpp
- gemm_aquant_pipeline_ag_bg_cr_v3.hpp
- gemm_bquant_pipeline_ag_bg_cr_base.hpp
- gemm_bquant_pipeline_ag_bg_cr_policy.hpp
- gemm_bquant_pipeline_ag_bg_cr_v3.hpp
- gemm_group_quant_utils.hpp
- gemm_microscale_pipeline_ag_bg_cr_base.hpp
- gemm_microscale_pipeline_ag_bg_cr_policy.hpp
- gemm_microscale_pipeline_ag_bg_cr_v3.hpp
- gemm_quant_pipeline_problem.hpp
- gemm_wp_abquant_pipeline_ag_bg_cr_base_policy.hpp
- gemm_wp_abquant_pipeline_ag_bg_cr_v2.hpp
- gemm_wp_bquant_pipeline_ag_bg_cr_base_policy.hpp
- gemm_wp_bquant_pipeline_ag_bg_cr_v2.hpp
- tile_gemm_quant_traits.hpp
- block/
- gemm_quant.hpp
- grouped_convolution/
- kernel/
- grouped_convolution_backward_data_kernel.hpp
- grouped_convolution_backward_weight_kernel.hpp
- grouped_convolution_forward_kernel.hpp
- pipeline/
- grouped_conv_universal_pipeline_ag_bg_cr_policy.hpp
- grouped_convolution_forward_depthwise_pipeline.hpp
- utils/
- convolution_specialization.hpp
- grouped_convolution_utils.hpp
- split_k_utils.hpp
- transform_conv_bwd_data_to_gemm.hpp
- transform_conv_bwd_weight_to_gemm.hpp
- transform_conv_fwd_to_gemm.hpp
- kernel/
- grouped_convolution.hpp
- image_to_column/
- kernel/
- image_to_column_kernel.hpp
- pipeline/
- block_image_to_column_problem.hpp
- tile_image_to_column_shape.hpp
- kernel/
- image_to_column.hpp
- layernorm2d/
- kernel/
- layernorm2d_fwd_kernel.hpp
- pipeline/
- layernorm2d_fwd_pipeline_default_policy.hpp
- layernorm2d_fwd_pipeline_one_pass.hpp
- layernorm2d_fwd_pipeline_problem.hpp
- layernorm2d_fwd_pipeline_two_pass.hpp
- layernorm2d_fwd_traits.hpp
- kernel/
- layernorm2d.hpp
- mask.hpp
- moe_flatmm.hpp
- norm_reduce/
- block/
- block_norm_reduce.hpp
- block_norm_reduce_problem.hpp
- thread/
- thread_welford.hpp
- block/
- norm_reduce.hpp
- permute/
- kernel/
- generic_permute_kernel.hpp
- pipeline/
- generic_petmute_problem.hpp
- kernel/
- permute.hpp
- pooling/
- kernel/
- pool_kernel.hpp
- pipeline/
- pool_default_policy.hpp
- pool_problem.hpp
- pool_shape.hpp
- kernel/
- pooling.hpp
- quant.hpp
- reduce/
- block/
- block_reduce.hpp
- block_reduce2d.hpp
- block_reduce2d_problem.hpp
- kernel/
- multi_reduce2d_kernel.hpp
- multi_reduce2d_multiblock_kernel.hpp
- multi_reduce2d_threadwise_kernel.hpp
- multi_reduce2d_tile_partitioner.hpp
- reduce2d_kernel.hpp
- pipeline/
- reduce2d_default_policy.hpp
- reduce2d_problem.hpp
- reduce2d_shape.hpp
- block/
- reduce.hpp
- rmsnorm2d/
- kernel/
- rmsnorm2d_fwd_kernel.hpp
- pipeline/
- rmsnorm2d_fwd_pipeline_default_policy.hpp
- rmsnorm2d_fwd_pipeline_model_sensitive_pass.hpp
- rmsnorm2d_fwd_pipeline_one_pass.hpp
- rmsnorm2d_fwd_pipeline_problem.hpp
- rmsnorm2d_fwd_pipeline_two_pass.hpp
- rmsnorm2d_fwd_traits.hpp
- kernel/
- rmsnorm2d.hpp
- rotary.hpp
- sageattention/
- block/
- block_sageattention_quant_scale_enum.hpp
- kernel/
- sageattn_fwd_kernel.hpp
- pipeline/
- block_sageattn_pipeline_enum.hpp
- block_sageattn_pipeline_problem.hpp
- block_sageattn_pipeline_qr_ks_vs.hpp
- block_sageattn_pipeline_qr_ks_vs_async.hpp
- block_sageattn_pipeline_qr_ks_vs_async_default_policy.hpp
- block_sageattn_pipeline_qr_ks_vs_custom_policy.hpp
- block_sageattn_pipeline_qr_ks_vs_default_policy.hpp
- tile_sageattn_shape.hpp
- tile_sageattn_traits.hpp
- block/
- sageattention.hpp
- sageattn.hpp
- smoothquant/
- kernel/
- moe_smoothquant_kernel.hpp
- smoothquant_kernel.hpp
- pipeline/
- smoothquant_pipeline_default_policy.hpp
- smoothquant_pipeline_one_pass.hpp
- smoothquant_pipeline_problem.hpp
- smoothquant_pipeline_two_pass.hpp
- kernel/
- smoothquant.hpp
- softmax/
- block/
- block_softmax_2d.hpp
- block_softmax_2d_problem.hpp
- block/
- softmax.hpp
- sparse_attn/
- kernel/
- fmha_fwd_jenga_kernel.hpp
- fmha_fwd_vsa_kernel.hpp
- pipeline/
- block_fmha_pipeline_qr_ks_vs_async_jenga.hpp
- block_fmha_pipeline_qr_ks_vs_async_vsa.hpp
- kernel/
- sparse_attn.hpp
- tdm/
- kernel/
- tdm_kernel.hpp
- kernel/
- tdm.hpp
- topk/
- block/
- block_topk_stream_2d.hpp
- block_topk_stream_2d_problem.hpp
- block/
- topk.hpp
- topk_softmax/
- kernel/
- topk_softmax_kernel.hpp
- pipeline/
- topk_softmax_warp_per_row_pipeline.hpp
- topk_softmax_warp_per_row_policy.hpp
- topk_softmax_warp_per_row_problem.hpp
- kernel/
- topk_softmax.hpp
- utils.hpp
- add_rmsnorm2d_rdquant/
- ref/
- README.md
- conv_common.hpp
- naive_attention.hpp
- naive_grouped_conv_bwd_data_gpu.hpp
- naive_grouped_conv_bwd_weight_gpu.hpp
- naive_grouped_conv_fwd_gpu.hpp
- remod.py
- utility/
- json_dump.hpp
- tile_load_store_microkernels.hpp
- ck/
- lib64/
- cmake/
- composable_kernel/
- composable_kernelConfig.cmake
- composable_kernelConfigVersion.cmake
- composable_kerneldevice_contraction_operationsTargets-relwithdebinfo.cmake
- composable_kerneldevice_contraction_operationsTargets.cmake
- composable_kerneldevice_conv_operationsTargets-relwithdebinfo.cmake
- composable_kerneldevice_conv_operationsTargets.cmake
- composable_kerneldevice_gemm_operationsTargets-relwithdebinfo.cmake
- composable_kerneldevice_gemm_operationsTargets.cmake
- composable_kerneldevice_other_operationsTargets-relwithdebinfo.cmake
- composable_kerneldevice_other_operationsTargets.cmake
- composable_kerneldevice_reduction_operationsTargets-relwithdebinfo.cmake
- composable_kerneldevice_reduction_operationsTargets.cmake
- composable_kernelutilityTargets-relwithdebinfo.cmake
- composable_kernelutilityTargets.cmake
- composable_kernel/
- libdevice_contraction_operations.so
- libdevice_conv_operations.so
- libdevice_gemm_operations.so
- libdevice_other_operations.so
- libdevice_reduction_operations.so
- libutility.so
- cmake/
- include/
Sources on Pagure