mirror of
https://github.com/ROCm/composable_kernel.git
synced 2026-05-04 21:51:28 +00:00
eb041079a3
* feat: grouped gemm tile loop support for RDNA4 * fix: removed extra parameter from grouped gemm example instance * fix: FP8 check incorrectly enabling FP8 on RDNA3
63 lines
1.9 KiB
C++
63 lines
1.9 KiB
C++
// Copyright (c) Advanced Micro Devices, Inc., or its affiliates.
|
|
// SPDX-License-Identifier: MIT
|
|
|
|
#include <vector>
|
|
|
|
#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
|
|
#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
|
|
#include "ck/tensor_operation/gpu/element/unary_element_wise_operation.hpp"
|
|
#include "ck/utility/tuple.hpp"
|
|
#include "ck/utility/data_type.hpp"
|
|
|
|
#include "gtest/gtest.h"
|
|
#include "test_grouped_gemm_util.hpp"
|
|
|
|
ck::index_t param_mask = 0xffffff;
|
|
ck::index_t instance_index = -1;
|
|
|
|
using F16 = ck::half_t;
|
|
using BF16 = ck::bhalf_t;
|
|
using F8 = ck::f8_t;
|
|
using I8 = int8_t;
|
|
|
|
using AElementOp = ck::tensor_operation::element_wise::PassThrough;
|
|
using BElementOp = ck::tensor_operation::element_wise::PassThrough;
|
|
using CDEElementOp = ck::tensor_operation::element_wise::FastGelu;
|
|
|
|
using Row = ck::tensor_layout::gemm::RowMajor;
|
|
using Col = ck::tensor_layout::gemm::ColumnMajor;
|
|
|
|
template <typename Tuple>
|
|
class TestGroupedGemm : public ck::test::TestGroupedGemm<Tuple, true>
|
|
{
|
|
};
|
|
|
|
// clang-format off
|
|
using KernelTypes = ::testing::Types<
|
|
ck::Tuple< Row, Row, Row, F16, F16, F16, AElementOp, BElementOp, CDEElementOp>,
|
|
ck::Tuple< Row, Col, Row, F16, F16, F16, AElementOp, BElementOp, CDEElementOp>,
|
|
ck::Tuple< Col, Row, Row, F16, F16, F16, AElementOp, BElementOp, CDEElementOp>,
|
|
ck::Tuple< Col, Col, Row, F16, F16, F16, AElementOp, BElementOp, CDEElementOp>
|
|
>;
|
|
// clang-format on
|
|
|
|
TYPED_TEST_SUITE(TestGroupedGemm, KernelTypes);
|
|
|
|
#include "test_grouped_gemm_ut_cases.inc"
|
|
int main(int argc, char** argv)
|
|
{
|
|
testing::InitGoogleTest(&argc, argv);
|
|
if(argc == 1) {}
|
|
else if(argc == 3)
|
|
{
|
|
param_mask = strtol(argv[1], nullptr, 0);
|
|
instance_index = atoi(argv[2]);
|
|
}
|
|
else
|
|
{
|
|
std::cout << "Usage of " << argv[0] << std::endl;
|
|
std::cout << "Arg1,2: param_mask instance_index(-1 means all)" << std::endl;
|
|
}
|
|
return RUN_ALL_TESTS();
|
|
}
|