Batchnorm inference instances, external API, client examples and gtests (#531)

* File renaming and class renaming for device element-wise operation * Add batchnorm-infer instances, external API and client example * Add batchnorm-infer profiler module and gtests * Remove file device_elementwise_extension.hpp and move NormalizeInInfer operation to element_wise_operation.hpp * Remove the using of class aliasing for DeviceElementwiseForBatchNormInfer * Rename class and file due to conflict from device_elementwise_2d.hpp * Fix namespace in batcnnorm_infer_nhwc client example
2026-04-20 06:49:15 +00:00 · 2023-01-26 07:09:04 +08:00
parent 52abc2f371
commit a1b2441f8d
29 changed files with 1260 additions and 80 deletions
--- a/test/batchnorm/batchnorm_infer_rank_4.cpp
+++ b/test/batchnorm/batchnorm_infer_rank_4.cpp
@@ -0,0 +1,89 @@
+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+#include <iostream>
+#include <initializer_list>
+#include <vector>
+#include <tuple>
+#include <gtest/gtest.h>
+
+#include "profiler/profile_batchnorm_infer_impl.hpp"
+
+using F16  = ck::half_t;
+using F32  = float;
+using BF16 = ck::bhalf_t;
+using F64  = double;
+
+template <typename Tuple>
+class TestBatchNormInferRank4 : public ::testing::Test
+{
+    private:
+    const double epsilon = std::numeric_limits<float>::epsilon();
+
+    protected:
+    using XDataType       = std::tuple_element_t<0, Tuple>;
+    using YDataType       = std::tuple_element_t<1, Tuple>;
+    using AccDataType     = std::tuple_element_t<2, Tuple>;
+    using ScaleDataType   = std::tuple_element_t<3, Tuple>;
+    using BiasDataType    = std::tuple_element_t<4, Tuple>;
+    using MeanVarDataType = std::tuple_element_t<5, Tuple>;
+
+    std::vector<std::vector<size_t>> list_of_lengths = {
+        {128, 16, 3, 1024}, {128, 16, 6, 512}, {4, 4, 4, 4}, {32, 32, 32, 32}};
+    std::vector<int> reduceDims;
+
+    template <int NumReduceDim>
+    void Run()
+    {
+        for(auto& inOutLengths : list_of_lengths)
+        {
+            bool pass = true;
+
+            EXPECT_FALSE(reduceDims.size() != NumReduceDim);
+
+            pass = pass && ck::profiler::profile_batchnorm_infer_impl<XDataType,
+                                                                      YDataType,
+                                                                      AccDataType,
+                                                                      ScaleDataType,
+                                                                      BiasDataType,
+                                                                      MeanVarDataType,
+                                                                      4,
+                                                                      NumReduceDim>(
+                               true, 3, false, false, inOutLengths, reduceDims, epsilon);
+
+            pass = pass && ck::profiler::profile_batchnorm_infer_impl<XDataType,
+                                                                      YDataType,
+                                                                      AccDataType,
+                                                                      ScaleDataType,
+                                                                      BiasDataType,
+                                                                      MeanVarDataType,
+                                                                      4,
+                                                                      NumReduceDim>(
+                               true, 3, false, false, inOutLengths, reduceDims, epsilon);
+
+            EXPECT_TRUE(pass);
+        }
+    }
+};
+
+using KernelTypes = ::testing::Types<std::tuple<F16, F16, F32, F16, F16, F32>,
+                                     std::tuple<F32, F32, F32, F32, F32, F32>,
+                                     std::tuple<BF16, BF16, F32, BF16, BF16, F32>,
+                                     std::tuple<F64, F64, F64, F64, F64, F64>>;
+
+TYPED_TEST_SUITE(TestBatchNormInferRank4, KernelTypes);
+
+// nhwc
+TYPED_TEST(TestBatchNormInferRank4, nhwc)
+{
+    this->reduceDims = {0, 1, 2};
+    this->template Run<3>();
+}
+
+// nchw
+TYPED_TEST(TestBatchNormInferRank4, nchw)
+{
+    this->reduceDims = {0, 2, 3};
+    this->template Run<3>();
+}