ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-15 18:42:06 +00:00
Code Issues Packages Projects Releases Wiki Activity

[CK] Fix misc issues in CK examples (#2890)

Browse Source 
* [CK] Fix misc CK issues

* revert fp8 change, it causes CI fail.

* resubmit fp8 change

[ROCm/composable_kernel commit: f076f207ce]
This commit is contained in:
linqunAMD
2025-09-25 02:28:20 +08:00
committed by GitHub
parent 7e537fd72f
commit 0c45597a4e
10 changed files with 74 additions and 79 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
example/01_gemm/gemm_xdl_skip_b_lds_fp16.cpp
View File

@@ -36,7 +36,7 @@ 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, 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, 4, 7, 1>;
< F32, F32, F32, F32, Row, Col, Row, PassThrough, PassThrough, PassThrough, GemmDefault, 256, 16, 128, 4, 4, 16, 16, 1, 2, S<16, 16, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 4, 4, true, 4, 4, 7, 1>;
using ADataType = float;
using BDataType = float;
using CDataType = float;
@@ -185,7 +185,6 @@ int main(int argc, char* argv[])
auto a_element_op = AElementOp{};
auto b_element_op = BElementOp{};
auto c_element_op = CElementOp{};
// do GEMM
auto gemm = DeviceGemmInstance{};
auto invoker = gemm.MakeInvoker();
@@ -209,8 +208,7 @@ int main(int argc, char* argv[])
return 0;
}
float ave_time = invoker.Run(argument, StreamConfig{nullptr, time_kernel});
float ave_time = invoker.Run(argument, StreamConfig{nullptr, time_kernel});
std::size_t flop = std::size_t(2) * M * N * K;
std::size_t num_btype =
sizeof(ADataType) * M * K + sizeof(BDataType) * K * N + sizeof(CDataType) * M * N;

2
example/10_convnd_fwd_multiple_d_multiple_reduce/common.hpp
View File

@@ -125,7 +125,7 @@ inline bool parse_cmd_args(int argc,
const ck::index_t num_dim_spatial = std::stoi(argv[4]);
problem_size = ck::utils::conv::parse_conv_param(
num_dim_spatial, threshold_to_catch_partial_args, argv);
num_dim_spatial, threshold_to_catch_partial_args + 1, argv);
}
else
{

2
example/10_convnd_fwd_multiple_d_multiple_reduce/run_convnd_fwd_max_example.inc
View File

@@ -23,7 +23,7 @@ using RsGlobalReduceOp =
static constexpr auto ConvSpec =
ck::tensor_operation::device::ConvolutionForwardSpecialization::Default;
static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::Default;
static constexpr auto GemmDefault = ck::tensor_operation::device::GemmSpecialization::MNKPadding;
// clang-format off
template <ck::index_t NDimSpatial>

6
example/21_gemm_layernorm/gemm_bias_relu_add_layernorm_xdl_welford_fp16.cpp
View File

@@ -65,7 +65,7 @@ using DeviceOpInstance = ck::tensor_operation::device::DeviceGemmMultipleDLayern
//######| | | | | Type| Type| Type| DataType| Type| Type| Type| Type| Type| Elementwise| Elementwise| Elementwise| Elementwise| Spacialization| Prefetch| Size| Block| Block| Block| | | XDL| XDL| Per| Per| ThreadCluster| ThreadCluster| SrcAccessOrder| SrcVectorDim| SrcScalar| DstScalar| AddExtraM| ThreadCluster| ThreadCluster| SrcAccessOrder| SrcVectorDim| SrcScalar| DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave| ThreadClusterLengths| ScalarPerVector| ThreadClusterLengths| ThreadSliceSize|
//######| | | | | | | | | | | | | | Operation| Operation| Operation| Operation| | Stage| | | | | | | | | Wave| Wave| Lengths_K0_M_K1| ArrangeOrder| | | PerVector| PerVector_K1| | Lengths_K0_N_K1| ArrangeOrder| | | PerVector| PerVector_K1| | PerShuffle| PerShuffle| _M_N| _M_N| _M_N| _M|
//######| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, AccDataType, CShuffleDataType, DsDataType, EMeanVarDataType, GammaDataType, BetaDataType, HDataType, AElementOp, BElementOp, CDEElementOp, HElementOp, GemmDefault, 1, 256, 256, 128, 32, 8, 8, 32, 32, 4, 2, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 8, 8, 1, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 8, 8, 1, 1, 1, S<32, 8>, 8, S<8, 32>, 8>;
< ALayout, BLayout, DsLayout, HLayout, ADataType, BDataType, AccDataType, CShuffleDataType, DsDataType, EMeanVarDataType, GammaDataType, BetaDataType, HDataType, AElementOp, BElementOp, CDEElementOp, HElementOp, GemmDefault, 1, 256, 256, 128, 32, 8, 8, 16, 16, 8, 4, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 8, 8, 1, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 8, 8, 1, 1, 1, S<32, 8>, 4, S<8, 32>, 4>;
// clang-format on
auto f_host_tensor_descriptor1d = [](std::size_t len, std::size_t stride) {
@@ -154,8 +154,8 @@ void host_gemm_layernorm(Tensor<HDataType>& h_m_n,
int main()
{
// temp disable on gfx11 & gfx12
if(ck::is_gfx11_supported() || ck::is_gfx12_supported())
// temp disable on gfx11
if(ck::is_gfx11_supported())
{
return 0;
}