ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-04-20 06:49:15 +00:00
Code Issues Packages Projects Releases Wiki Activity

Ck int4 moe develop (#1949)

Browse Source 
* Add Gemm fp8xint4 example and kernel, function pass.

* Init Gemm_fp8xint4 Bpreshuffle

* Added gemm_fp8xint4_Bpreshuffle files, function not checked yet

* General fix.

* fp8xint4 bpreshuffle function pass

* fix.

* init b preshuffle dequant in VGPR.

* fix bug, function pass.

* move b thread dequant copy to blockwise.

* fix bug, function now passes.

* modified the tile size to 256, 128x128x128.

* fixed a bug.

* Initial int4 moe, compile pass, function not check.

* fix bug in moe_gemm1.cpp, now function pass.

* test expert = 8 and function pass.

* Added moe_pk_i4_gemm2, function pass.

* Added b preshuffle pipeline v3 support.

* fixed merge issue. fp8xint4 and fp8xint4_bpreshuffle function pass.

* Split the blockwise pipeline for fp8xint4.

* commit missing files

* opt gemm2 to 2x2 wave

* fix swizzle = false

* update int4 moe with latest input changes.

* update tile size.

* enable pipeline v3.

* fix nswizzle = true

* commit a version for compiler debug.

* Updated transfer_v3r1_gather to support pk_i4_t type.

* for int4 moe2 for type_convert support.

* remove some values between mfma instructions.

* fix int4 moe

* Updated transfer_v3r1_gather to support pk_i4_t type.

* i4 support lds multiple shuffle

* fixed int4 moe tflops calculation.

* Modified CshuffleCShuffleMXdlPerWavePerShuffle to 1 to suit C multiple shuffle

* updated gemm2.

* change int4 moe example names

* fix and format code.

* format.

* format codes.

* update fp8xint4 example tile size.

* add <unordered_map> header

* fixed.

* format.

* Added conditional compilation for int4 -> fp8 conversion kernels

---------

Co-authored-by: mtgu0705 <mtgu@amd.com>
Co-authored-by: coderfeli <coderfeli@163.com>
This commit is contained in:
Mingtao Gu
2025-03-10 11:16:44 +08:00
committed by GitHub
parent c954bd0cfa
commit 0db7c8f0b2
19 changed files with 6018 additions and 83 deletions
Download Patch File Download Diff File Expand all files Collapse all files

23
example/01_gemm/common.hpp
View File

@@ -7,6 +7,7 @@
#include <iostream>
#include <initializer_list>
#include <numeric>
#include <unordered_map>
#include "ck/ck.hpp"
#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
@@ -369,3 +370,25 @@ inline __host__ __device__ constexpr double get_atol()
return 1e-3;
}
}
float i4_to_f32_gfx9(uint8_t i4)
{
static std::unordered_map<uint8_t, float> u = {{0b1000, -0.5000f},
{0b1001, -0.4375f},
{0b1010, -0.3750f},
{0b1011, -0.3125f},
{0b1100, -0.2500f},
{0b1101, -0.1875f},
{0b1110, -0.1250f},
{0b1111, -0.0625f},
{0b0, +0.0000f},
{0b1, +0.0625f},
{0b10, +0.1250f},
{0b11, +0.1875f},
{0b100, +0.2500f},
{0b101, +0.3125f},
{0b110, +0.3750f},
{0b111, +0.4375f}};
return u[i4];
}