Skip to content

GitLab

Commit 179c561e authored by ltqin's avatar ltqin
Browse files

fix some config

parent 428ae72a
Hide whitespace changes
Inline Side-by-side
Showing with 10 additions and 10 deletions
example/01_gemm/gemm_xdl_skip_b_lds_fp16.cpp
View file @ 179c561e
......@@ -48,15 +48,15 @@ using DeviceGemmInstance = ck::tensor_operation::device::DeviceGemmXdlSkipBLds
//###########| | | | | | | | Operation| Operation| Operation| | | | | | | | | Wave| Wave| Lengths_K0_M_K1| ArrangeOrder| | | PerVector| PerVector_K1| | PerVector| | PerVector|
//###########| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
#if 0
< F16, F16, F16, F32, Row, Col, Row, PassThrough, PassThrough, PassThrough, GemmDefault, 256, 64, 128, 4, 8, 32, 32, 1, 2, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 8, 8, true, 8, 7, 1>;
< F16, F16, F16, F32, Row, Col, Row, PassThrough, PassThrough, PassThrough, GemmDefault, 256, 32, 128, 4, 8, 32, 32, 1, 1, S<8, 32, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 8, 8, true, 8, 7, 1>;
using ADataType = ck::half_t;
using BDataType = ck::half_t;
using CDataType = ck::half_t;
using AccDataType = float;
#else
// < F32, F32, F32, F32, Row, Col, Row, PassThrough, PassThrough, PassThrough, GemmDefault, 256, 64, 16, 4, 4, 16, 16, 1, 1, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 4, 4, true, 4, 7, 1>;
< F32, F32, F32, F32, Row, Col, Row, PassThrough, PassThrough, PassThrough, GemmDefault, 256, 16, 64, 4, 4, 16, 16, 1, 1, S<16, 16, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 4, 4, true, 4, 7, 1>;
// < F32, F32, F32, F32, Row, Col, Row, PassThrough, PassThrough, PassThrough, GemmDefault, 256, 128, 32, 4, 4, 32, 32, 1, 1, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 4, 4, true, 4, 7, 1>;
< F32, F32, F32, F32, Row, Col, Row, PassThrough, PassThrough, PassThrough, GemmDefault, 256, 128, 64, 4, 4, 32, 32, 2, 1, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 4, 4, true, 4, 7, 1>;
// < F32, F32, F32, F32, Row, Col, Row, PassThrough, PassThrough, PassThrough, GemmDefault, 256, 128, 64, 4, 4, 32, 32, 2, 1, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 4, 4, true, 4, 7, 1>;
using ADataType = float;
using BDataType = float;
using CDataType = float;
......@@ -69,9 +69,9 @@ using DeviceGemmInstance = ck::tensor_operation::device::DeviceGemmXdl
//###########| Type| Type| Type| Type| | | | Elementwise| Elementwise| Elementwise|Spacialization| Size| Block| Block| Block| | XDL| XDL| Per| Per| ThreadCluster| ThreadCluster| SrcAccessOrder| SrcVectorDim| SrcScalar| DstScalar| AddExtraM| ThreadCluster| ThreadCluster| SrcAccessOrder| SrcVectorDim| SrcScalar| DstScalar| AddExtraN| SrcDstVectorDim| DstScalar|
//###########| | | | | | | | Operation| Operation| Operation| | | | | | | | | Wave| Wave| Lengths_K0_M_K1| ArrangeOrder| | | PerVector| PerVector_K1| | Lengths_K0_N_K1| ArrangeOrder| | | PerVector| PerVector_K1| | | PerVector|
//###########| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
// < F32, F32, F32, F32, Row, Col, Row, PassThrough, PassThrough, PassThrough, GemmDefault, 256, 64, 16, 4, 4, 16, 16, 1, 1, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 4, 4, true, S<4, 16, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 4, 4, true, 7, 1, 2>;
< F32, F32, F32, F32, Row, Col, Row, PassThrough, PassThrough, PassThrough, GemmDefault, 256, 16, 64, 4, 4, 16, 16, 1, 1, S<4, 16, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 4, 4, true, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 4, 4, true, 7, 1, 2>;
// < F32, F32, F32, F32, Row, Col, Row, PassThrough, PassThrough, PassThrough, GemmDefault, 256, 128, 32, 4, 4, 32, 32, 1, 1, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 4, 4, true, S<4, 32, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 4, 4, true, 7, 1, 2>;
< F32, F32, F32, F32, Row, Col, Row, PassThrough, PassThrough, PassThrough, GemmDefault, 256, 128, 64, 4, 4, 32, 32, 1, 2, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 4, 4, true, S<4, 16, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 4, 4, true, 7, 1, 2>;
// < F32, F32, F32, F32, Row, Col, Row, PassThrough, PassThrough, PassThrough, GemmDefault, 256, 128, 64, 4, 4, 32, 32, 1, 2, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 4, 4, true, S<4, 16, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 4, 4, true, 7, 1, 2>;
using ADataType = float;
using BDataType = float;
using CDataType = float;
......@@ -107,12 +107,12 @@ int main(int argc, char* argv[])
// GEMM shape
#if 1
ck::index_t M = 128 * 360;
ck::index_t N = 64;
ck::index_t M = 16;
ck::index_t N = 64 * 720;
ck::index_t K = 4096;
ck::index_t StrideA = 4096;
ck::index_t StrideB = 4096;
ck::index_t StrideA = K;
ck::index_t StrideB = K;
ck::index_t StrideC = N;
#else
ck::index_t M = 16;
......
include/ck/tensor_operation/gpu/grid/gridwise_gemm_xdlops_skip_b_lds_v1.hpp
View file @ 179c561e
......@@ -112,7 +112,7 @@ struct GridwiseGemm_k0mk1_k0nk1_mn_xdlops_skip_b_lds_v1
static constexpr auto I6 = Number<6>{};
static constexpr auto I7 = Number<7>{};
static constexpr auto MultiK0 = 2 * 1;
static constexpr auto MultiK0 = 2 * 2;
// K1 should be Number<...>
static constexpr auto K1 = Number<K1Value>{};
......
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