Files
composable_kernel/test/batched_gemm/test_batched_gemm_xdl.cpp
Illia Silin ae57e5938e Split the instances by architecture. (#1223)
* parse examples inside the add_example_executable function

* fix the example 64 cmake file

* add xdl flag to the gemm_bias_softmax_gemm_permute example

* add filtering of tests based on architecture type

* enable test_grouped_gemm for gfx9 only

* enable test_transpose only for gfx9

* only linnk test_transpose if it gets built

* split the gemm instances by architectures

* split gemm_bilinear,grouped_conv_bwd_weight instances by targets

* split instances by architecture

* split grouped_conv instances by architecture

* fix clang format

* fix the if-else logic in group_conv headers

* small fix for grouped convolution instances

* fix the grouped conv bwd weight dl instances

* fix client examples

* only enable client examples 3 and 4 on gfx9

* set the gfx9 macro

* make sure the architecture macros are set by cmake

* use separate set of xdl/wmma flags for host code

* sinmplify the main cmake file

* add conv_fwd_bf8 instance declaration
2024-04-02 09:42:17 -07:00

186 lines
8.8 KiB
C++

// SPDX-License-Identifier: MIT
// Copyright (c) 2023, Advanced Micro Devices, Inc. All rights reserved.
#include <cstdlib>
#include <iostream>
#include <initializer_list>
#include <tuple>
#include <vector>
#include <gtest/gtest.h>
#include "profiler/profile_batched_gemm_impl.hpp"
#include "ck/library/tensor_operation_instance/gpu/batched_gemm.hpp"
struct GemmParams
{
ck::index_t M;
ck::index_t N;
ck::index_t K;
ck::index_t BatchCount;
};
class TestBatchedGemm : public ::testing::Test
{
protected:
using Row = ck::tensor_layout::gemm::RowMajor;
using Col = ck::tensor_layout::gemm::ColumnMajor;
using PassThrough = ck::tensor_operation::element_wise::PassThrough;
std::vector<GemmParams> params;
template <typename DataType>
void Run()
{
using namespace ck::tensor_operation::device;
bool pass = true;
for(auto& param : params)
{
const auto M = param.M;
const auto N = param.N;
const auto K = param.K;
const auto BatchCount = param.BatchCount;
pass =
pass && ck::profiler::profile_batched_gemm_impl<DataType,
DataType,
DataType,
Row,
Row,
Row,
PassThrough,
PassThrough,
PassThrough,
DeviceBatchedGemm<Row,
Row,
Row,
DataType,
DataType,
DataType,
PassThrough,
PassThrough,
PassThrough>>(
true, 1, false, 1, M, N, K, K, N, N, M * K, K * N, M * N, BatchCount);
pass =
pass && ck::profiler::profile_batched_gemm_impl<DataType,
DataType,
DataType,
Row,
Col,
Row,
PassThrough,
PassThrough,
PassThrough,
DeviceBatchedGemm<Row,
Col,
Row,
DataType,
DataType,
DataType,
PassThrough,
PassThrough,
PassThrough>>(
true, 1, false, 1, M, N, K, K, K, N, M * K, K * N, M * N, BatchCount);
pass =
pass && ck::profiler::profile_batched_gemm_impl<DataType,
DataType,
DataType,
Col,
Row,
Row,
PassThrough,
PassThrough,
PassThrough,
DeviceBatchedGemm<Col,
Row,
Row,
DataType,
DataType,
DataType,
PassThrough,
PassThrough,
PassThrough>>(
true, 1, false, 1, M, N, K, M, N, N, M * K, K * N, M * N, BatchCount);
pass =
pass && ck::profiler::profile_batched_gemm_impl<DataType,
DataType,
DataType,
Col,
Col,
Row,
PassThrough,
PassThrough,
PassThrough,
DeviceBatchedGemm<Col,
Col,
Row,
DataType,
DataType,
DataType,
PassThrough,
PassThrough,
PassThrough>>(
true, 1, false, 1, M, N, K, M, K, N, M * K, K * N, M * N, BatchCount);
}
EXPECT_TRUE(pass);
}
};
#ifdef CK_ENABLE_INT8
TEST_F(TestBatchedGemm, i8)
{
this->params.push_back({64, 64, 64, 2});
this->params.push_back({64, 64, 64, 1});
this->params.push_back({60, 60, 60, 2});
this->params.push_back({68, 68, 68, 2});
this->params.push_back({40, 40, 40, 2});
this->params.push_back({256, 256, 128, 3});
this->template Run<int8_t>();
}
#endif
#ifdef CK_ENABLE_BF16
TEST_F(TestBatchedGemm, bf16)
{
this->params.push_back({64, 64, 64, 2});
this->params.push_back({64, 64, 64, 1});
this->params.push_back({60, 60, 60, 2});
this->params.push_back({68, 68, 68, 2});
this->params.push_back({40, 40, 40, 2});
this->params.push_back({256, 256, 128, 3});
this->template Run<ck::bhalf_t>();
}
#endif
#ifdef CK_ENABLE_FP16
TEST_F(TestBatchedGemm, fp16)
{
this->params.push_back({64, 64, 64, 2});
this->params.push_back({64, 64, 64, 1});
this->params.push_back({60, 60, 60, 2});
this->params.push_back({68, 68, 68, 2});
this->params.push_back({40, 40, 40, 2});
this->params.push_back({256, 256, 128, 3});
this->template Run<ck::half_t>();
}
#endif
#ifdef CK_ENABLE_FP32
TEST_F(TestBatchedGemm, fp32)
{
this->params.push_back({64, 64, 64, 2});
this->params.push_back({64, 64, 64, 1});
this->params.push_back({60, 60, 60, 2});
this->params.push_back({68, 68, 68, 2});
this->params.push_back({40, 40, 40, 2});
this->params.push_back({256, 256, 128, 3});
this->template Run<float>();
}
#endif