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

add int8 gemm multiply multiply a8w8 (#1591)

Browse Source 
* add int8 gemm multiply multiply a8w8

* uncomment

* clang-format-12

* Add example_gemm_multiply_multiply_xdl_int8

* Remove shell scripts

* update preprocess number for mi308; bring back printout in ckprofiler

* format

---------

Co-authored-by: chenjun <junchen2@amd.com>
Co-authored-by: Haocong WANG <haocwang@amd.com>
Co-authored-by: carlushuang <carlus.huang@amd.com>
This commit is contained in:
valarLip
2024-10-26 16:39:34 +08:00
committed by GitHub
parent eda5938386
commit 37f7afed1e
16 changed files with 794 additions and 28 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
include/ck/tensor_operation/gpu/element/element_wise_operation.hpp
View File

@@ -272,6 +272,26 @@ struct MultiplyMultiply
e = ck::type_convert<ck::bhalf_t>(x0_f);
}
template <>
__host__ __device__ constexpr void operator()<ck::half_t, int, ck::half_t, ck::half_t>(
ck::half_t& e, const int& c, const ck::half_t& d0, const ck::half_t& d1) const
{
const float x0_f =
ck::type_convert<float>(c) * ck::type_convert<float>(d0) * ck::type_convert<float>(d1);
e = ck::type_convert<ck::half_t>(x0_f);
}
template <>
__host__ __device__ constexpr void operator()<ck::bhalf_t, int, float, float>(
ck::bhalf_t& e, const int& c, const float& d0, const float& d1) const
{
const float x0_f =
ck::type_convert<float>(c) * ck::type_convert<float>(d0) * ck::type_convert<float>(d1);
e = ck::type_convert<ck::bhalf_t>(x0_f);
}
};
struct MultiplyAddFastGelu