Skip to content

GitLab

Commit a55440c0 authored by Bartlomiej Wroblewski's avatar Bartlomiej Wroblewski
Browse files

Review: Rename DppInstrRunner to DppLanegroupGemm

parent 792b62dd
Hide whitespace changes
Inline Side-by-side
Showing with 25 additions and 25 deletions
include/ck/tensor_operation/gpu/warp/dpp_gemm.hpp
View file @ a55440c0
...@@ -59,14 +59,14 @@ struct dpp_type<DppInstr::dpp8_f16_32x8x2> ...@@ -59,14 +59,14 @@ struct dpp_type<DppInstr::dpp8_f16_32x8x2>
template <index_t MPerDpp, index_t NPerDpp, class ADataType, class BDataType, class CDataType> template <index_t MPerDpp, index_t NPerDpp, class ADataType, class BDataType, class CDataType>
__device__ void run(const ADataType& a, const BDataType& b, CDataType& reg_c) const __device__ void run(const ADataType& a, const BDataType& b, CDataType& reg_c) const
{ {
dpp8::DppInstrRunner<m_per_thread, dpp8::DppLanegroupGemm<m_per_thread,
n_per_thread, n_per_thread,
k_per_dpp, k_per_dpp,
BaseType, BaseType,
ADataType, ADataType,
BDataType, BDataType,
CDataType, CDataType,
share_a>{} share_a>{}
.Run(a, b, reg_c); .Run(a, b, reg_c);
} }
}; };
...@@ -89,14 +89,14 @@ struct dpp_type<DppInstr::dpp8_f16_8x32x2> ...@@ -89,14 +89,14 @@ struct dpp_type<DppInstr::dpp8_f16_8x32x2>
template <index_t MPerDpp, index_t NPerDpp, class ADataType, class BDataType, class CDataType> template <index_t MPerDpp, index_t NPerDpp, class ADataType, class BDataType, class CDataType>
__device__ void run(const ADataType& a, const BDataType& b, CDataType& reg_c) const __device__ void run(const ADataType& a, const BDataType& b, CDataType& reg_c) const
{ {
dpp8::DppInstrRunner<m_per_thread, dpp8::DppLanegroupGemm<m_per_thread,
n_per_thread, n_per_thread,
k_per_dpp, k_per_dpp,
BaseType, BaseType,
ADataType, ADataType,
BDataType, BDataType,
CDataType, CDataType,
share_a>{} share_a>{}
.Run(a, b, reg_c); .Run(a, b, reg_c);
} }
}; };
...@@ -119,14 +119,14 @@ struct dpp_type<DppInstr::dpp8_f16_16x16x2> ...@@ -119,14 +119,14 @@ struct dpp_type<DppInstr::dpp8_f16_16x16x2>
template <index_t MPerDpp, index_t NPerDpp, class ADataType, class BDataType, class CDataType> template <index_t MPerDpp, index_t NPerDpp, class ADataType, class BDataType, class CDataType>
__device__ void run(const ADataType& a, const BDataType& b, CDataType& reg_c) const __device__ void run(const ADataType& a, const BDataType& b, CDataType& reg_c) const
{ {
dpp8::DppInstrRunner<m_per_thread, dpp8::DppLanegroupGemm<m_per_thread,
n_per_thread, n_per_thread,
k_per_dpp, k_per_dpp,
BaseType, BaseType,
ADataType, ADataType,
BDataType, BDataType,
CDataType, CDataType,
share_a>{} share_a>{}
.Run(a, b, reg_c); .Run(a, b, reg_c);
} }
}; };
......
include/ck/utility/amd_gemm_dpp.hpp
View file @ a55440c0
...@@ -33,7 +33,7 @@ template <index_t MPerThread, ...@@ -33,7 +33,7 @@ template <index_t MPerThread,
class BVecDataType, class BVecDataType,
class CVecDataType, class CVecDataType,
bool ShareA> bool ShareA>
struct DppInstrRunner struct DppLanegroupGemm
{ {
using datatypes_conf = dpp_datatypes<BaseInputType>; using datatypes_conf = dpp_datatypes<BaseInputType>;
using ADataType = typename datatypes_conf::a_dtype; using ADataType = typename datatypes_conf::a_dtype;
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment