Skip to content

GitLab

Commit 747039cc authored by Jing Zhang's avatar Jing Zhang
Browse files

adjust tuning params

parent a422df82
Show whitespace changes
Inline Side-by-side
Showing with 1 addition and 1 deletion
example/01_gemm/gemm_dl_fp16.cpp
View file @ 747039cc
...@@ -27,8 +27,8 @@ using DeviceGemmInstance = ck::tensor_operation::device::DeviceGemmDl ...@@ -27,8 +27,8 @@ using DeviceGemmInstance = ck::tensor_operation::device::DeviceGemmDl
// ######| Type| Type| Type| Type| | | | Elementwise| Elementwise| Elementwise| Spacialization| Size| Block| Block| Block| | ThreadM111| ThreadN111| Thread| ThreadSliceLengths| ThreadClusterLengths| ThreadCluster| SrcAccess| SrcVectorTensor| SrcVectorTensor| DstVectorTensor| SrcAccess| SrcVectorDim| SrcScalarPerVector| SrcDstAccess| SrcDstVectorDim| DstScalarPerVector| // ######| Type| Type| Type| Type| | | | Elementwise| Elementwise| Elementwise| Spacialization| Size| Block| Block| Block| | ThreadM111| ThreadN111| Thread| ThreadSliceLengths| ThreadClusterLengths| ThreadCluster| SrcAccess| SrcVectorTensor| SrcVectorTensor| DstVectorTensor| SrcAccess| SrcVectorDim| SrcScalarPerVector| SrcDstAccess| SrcDstVectorDim| DstScalarPerVector|
// ######| | | | | | | | Operation| Operation| Operation| | | | | | | | | | K0_M0_M1_K1| K0_M0_M1_K1| ArrangeOrder| Order| Lengths_K0_M0_M1_K1| ContiguousDimOrder| Lengths_K0_M0_M1_K1| Order| | | Order| | | // ######| | | | | | | | Operation| Operation| Operation| | | | | | | | | | K0_M0_M1_K1| K0_M0_M1_K1| ArrangeOrder| Order| Lengths_K0_M0_M1_K1| ContiguousDimOrder| Lengths_K0_M0_M1_K1| Order| | | Order| | |
// ######| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | // ######| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementOp, BElementOp, CElementOp, GemmMNPadding, 64, 4, 64, 16, 2, 1, 4, 1, S<1, 1, 1, 2>, S<16, 1, 4, 1>, S<1, 2, 0, 3>, S<1, 2, 0, 3>, S<1, 1, 1, 2>, S<1, 2, 0, 3>, S<1, 1, 1, 2>, S<1, 2, 0, 3>, 3, 2, S<0, 1, 2, 3, 4, 5>, 5, 4>;
//< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementOp, BElementOp, CElementOp, GemmMNPadding, 64, 1, 64, 32, 2, 1, 1, 1, S<1, 1, 1, 2>, S<32, 1, 1, 1>, S<1, 2, 0, 3>, S<1, 2, 0, 3>, S<1, 1, 1, 2>, S<1, 2, 0, 3>, S<1, 1, 1, 2>, S<1, 2, 0, 3>, 3, 2, S<0, 1, 2, 3, 4, 5>, 5, 1>; //< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementOp, BElementOp, CElementOp, GemmMNPadding, 64, 1, 64, 32, 2, 1, 1, 1, S<1, 1, 1, 2>, S<32, 1, 1, 1>, S<1, 2, 0, 3>, S<1, 2, 0, 3>, S<1, 1, 1, 2>, S<1, 2, 0, 3>, S<1, 1, 1, 2>, S<1, 2, 0, 3>, 3, 2, S<0, 1, 2, 3, 4, 5>, 5, 1>;
< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementOp, BElementOp, CElementOp, GemmMNPadding, 64, 1, 64, 16, 4, 1, 1, 1, S<1, 1, 1, 4>, S<16, 1, 1, 1>, S<1, 2, 0, 3>, S<1, 2, 0, 3>, S<1, 1, 1, 4>, S<1, 2, 0, 3>, S<1, 1, 1, 2>, S<1, 2, 0, 3>, 3, 4, S<0, 1, 2, 3, 4, 5>, 5, 1>;
// clang-format on // clang-format on
using ReferenceGemmInstance = ck::tensor_operation::host:: using ReferenceGemmInstance = ck::tensor_operation::host::
......
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