mirror of
https://github.com/ROCm/composable_kernel.git
synced 2026-06-28 18:56:59 +00:00
bdd7a8333d
[CK Tile] PermuteN support MX GEMM ## Motivation Add PermuteN support to preshuffle MX GEMM ## Technical Details - Modify `shuffle_b_permuteN` to support MX preshuffled layout - Add `preShuffleScalePermuteN` with same functionality of `preShuffleScale` but layout consistent with PermuteN - Include MX pre-processing functions in the library ## Test Plan Add test configuration for permuteN with preshuffle (both FP4 and FP8) ## Submission Checklist - [x] Look over the contributing guidelines at https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests. Co-authored-by: Cong Ma <congma13@amd.com>
58 lines
2.9 KiB
C++
58 lines
2.9 KiB
C++
// Copyright (c) Advanced Micro Devices, Inc., or its affiliates.
|
|
// SPDX-License-Identifier: MIT
|
|
#pragma once
|
|
|
|
#include "ck_tile/host/arg_parser.hpp"
|
|
#include "ck_tile/host/check_err.hpp"
|
|
#include "ck_tile/host/concat.hpp"
|
|
#include "ck_tile/host/convolution_host_tensor_descriptor_helper.hpp"
|
|
#include "ck_tile/host/convolution_parameter.hpp"
|
|
#include "ck_tile/host/device_memory.hpp"
|
|
#include "ck_tile/host/device_prop.hpp"
|
|
#include "ck_tile/host/fill.hpp"
|
|
#include "ck_tile/host/flush_icache.hpp"
|
|
#include "ck_tile/host/high_res_cpu_clock.hpp"
|
|
#include "ck_tile/host/hip_check_error.hpp"
|
|
#include "ck_tile/host/host_tensor.hpp"
|
|
#include "ck_tile/host/host_tensor_io.hpp"
|
|
#include "ck_tile/host/joinable_thread.hpp"
|
|
#include "ck_tile/host/kernel_launch.hpp"
|
|
#include "ck_tile/host/mx_processing.hpp"
|
|
#include "ck_tile/host/permute_pk_int4.hpp"
|
|
#include "ck_tile/host/pinned_host_releaser.hpp"
|
|
#include "ck_tile/host/ranges.hpp"
|
|
#include "ck_tile/host/reference/reference_batched_contraction.hpp"
|
|
#include "ck_tile/host/reference/reference_batched_dropout.hpp"
|
|
#include "ck_tile/host/reference/reference_batched_dropout_randval.hpp"
|
|
#include "ck_tile/host/reference/reference_batched_elementwise.hpp"
|
|
#include "ck_tile/host/reference/reference_batched_gemm.hpp"
|
|
#include "ck_tile/host/reference/reference_batched_masking.hpp"
|
|
#include "ck_tile/host/reference/reference_batched_mx_descale.hpp"
|
|
#include "ck_tile/host/reference/reference_batched_rotary_position_embedding.hpp"
|
|
#include "ck_tile/host/reference/reference_batched_softmax.hpp"
|
|
#include "ck_tile/host/reference/reference_batched_transpose.hpp"
|
|
#include "ck_tile/host/reference/reference_blocked_attention.hpp"
|
|
#include "ck_tile/host/reference/reference_elementwise.hpp"
|
|
#include "ck_tile/host/reference/reference_fused_moe.hpp"
|
|
#include "ck_tile/host/reference/reference_gemm.hpp"
|
|
#include "ck_tile/host/reference/reference_grouped_conv_bwd_data.hpp"
|
|
#include "ck_tile/host/reference/reference_grouped_conv_bwd_weight.hpp"
|
|
#include "ck_tile/host/reference/reference_grouped_conv_fwd.hpp"
|
|
#include "ck_tile/host/reference/reference_im2col.hpp"
|
|
#include "ck_tile/host/reference/reference_layernorm2d_fwd.hpp"
|
|
#include "ck_tile/host/reference/reference_moe_gemm.hpp"
|
|
#include "ck_tile/host/reference/reference_moe_sorting.hpp"
|
|
#include "ck_tile/host/reference/reference_permute.hpp"
|
|
#include "ck_tile/host/reference/reference_pool.hpp"
|
|
#include "ck_tile/host/reference/reference_reduce.hpp"
|
|
#include "ck_tile/host/reference/reference_rmsnorm2d_fwd.hpp"
|
|
#include "ck_tile/host/reference/reference_rowwise_quantization2d.hpp"
|
|
#include "ck_tile/host/reference/reference_softmax.hpp"
|
|
#include "ck_tile/host/reference/reference_topk.hpp"
|
|
#include "ck_tile/host/reference/reference_transpose.hpp"
|
|
#include "ck_tile/host/rotating_buffers.hpp"
|
|
#include "ck_tile/host/stream_config.hpp"
|
|
#include "ck_tile/host/stream_utils.hpp"
|
|
#include "ck_tile/host/tensor_shuffle_utils.hpp"
|
|
#include "ck_tile/host/timer.hpp"
|