Add fp32 and bf16 tests (#487)

test/batched_gemm/CMakeLists.txt

@@ -2,6 +2,14 @@ add_test_executable(test_batched_gemm_fp16 batched_gemm_fp16.cpp)
 target_link_libraries(test_batched_gemm_fp16 PRIVATE utility)
 target_link_libraries(test_batched_gemm_fp16 PRIVATE device_batched_gemm_instance)
 
+add_test_executable(test_batched_gemm_fp32 batched_gemm_fp32.cpp)
+target_link_libraries(test_batched_gemm_fp32 PRIVATE utility)
+target_link_libraries(test_batched_gemm_fp32 PRIVATE device_batched_gemm_instance)
+
+add_test_executable(test_batched_gemm_bf16 batched_gemm_bf16.cpp)
+target_link_libraries(test_batched_gemm_bf16 PRIVATE utility)
+target_link_libraries(test_batched_gemm_bf16 PRIVATE device_batched_gemm_instance)
+
 add_test_executable(test_batched_gemm_int8 batched_gemm_int8.cpp)
 target_link_libraries(test_batched_gemm_int8 PRIVATE utility)
 target_link_libraries(test_batched_gemm_int8 PRIVATE device_batched_gemm_instance)

test/batched_gemm/batched_gemm_bf16.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <iostream>
+
+#include "profiler/include/profile_batched_gemm_impl.hpp"
+
+namespace {
+using ADataType = ck::bhalf_t;
+using BDataType = ck::bhalf_t;
+using CDataType = ck::bhalf_t;
+
+using Row = ck::tensor_layout::gemm::RowMajor;
+using Col = ck::tensor_layout::gemm::ColumnMajor;
+} // namespace
+
+int main()
+{
+    int M          = 256;
+    int N          = 256;
+    int K          = 128;
+    int BatchCount = 3;
+
+    bool pass = true;
+
+    pass = pass &&
+           ck::profiler::profile_batched_gemm_impl<ADataType, BDataType, CDataType, Row, Row, Row>(
+               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<ADataType, BDataType, CDataType, Row, Col, Row>(
+               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<ADataType, BDataType, CDataType, Col, Row, Row>(
+               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<ADataType, BDataType, CDataType, Col, Col, Row>(
+               true, 1, false, 1, M, N, K, M, K, N, M * K, K * N, M * N, BatchCount);
+
+    std::cout << "test BatchedGEMM bf16: " << (pass ? "Pass" : "Fail") << std::endl;
+    return pass ? 0 : 1;
+}

test/batched_gemm/batched_gemm_fp32.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <iostream>
+
+#include "profiler/include/profile_batched_gemm_impl.hpp"
+
+namespace {
+using ADataType = float;
+using BDataType = float;
+using CDataType = float;
+
+using Row = ck::tensor_layout::gemm::RowMajor;
+using Col = ck::tensor_layout::gemm::ColumnMajor;
+} // namespace
+
+int main()
+{
+    int M          = 256;
+    int N          = 256;
+    int K          = 128;
+    int BatchCount = 3;
+
+    bool pass = true;
+
+    pass = pass &&
+           ck::profiler::profile_batched_gemm_impl<ADataType, BDataType, CDataType, Row, Row, Row>(
+               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<ADataType, BDataType, CDataType, Row, Col, Row>(
+               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<ADataType, BDataType, CDataType, Col, Row, Row>(
+               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<ADataType, BDataType, CDataType, Col, Col, Row>(
+               true, 1, false, 1, M, N, K, M, K, N, M * K, K * N, M * N, BatchCount);
+
+    std::cout << "test BatchedGEMM fp32: " << (pass ? "Pass" : "Fail") << std::endl;
+    return pass ? 0 : 1;
+}