SWDEV-1 - Merge github PRs to amd-staging

- https://github.com/ROCm/hip-tests/pull/119 - https://github.com/ROCm/hip-tests/pull/151 - https://github.com/ROCm/hip-tests/pull/57 - https://github.com/ROCm/hip-tests/pull/58 - https://github.com/ROCm/hip-tests/pull/59 - https://github.com/ROCm/hip-tests/pull/60 - https://github.com/ROCm/hip-tests/pull/99 - https://github.com/ROCm/hip-tests/pull/139 - https://github.com/ROCm/hip-tests/pull/152 - https://github.com/ROCm/hip-tests/pull/48 - https://github.com/ROCm/hip-tests/pull/54 - https://github.com/ROCm/hip-tests/pull/53 - https://github.com/ROCm/hip-tests/pull/24 - https://github.com/ROCm/hip-tests/pull/23 - https://github.com/ROCm/hip-tests/pull/22 - https://github.com/ROCm/hip-tests/pull/21 - https://github.com/ROCm/hip-tests/pull/20 - https://github.com/ROCm/hip-tests/pull/14 - https://github.com/ROCm/hip-tests/pull/8 Change-Id: I1eea54cd1436f3ddbfd5c1b3b2f672eb81d03cd4 [ROCm/hip-tests commit: 96df1fde80]
2024-02-22 18:31:56 +05:30
parent 4b6fb4fff4 d1bebf302f
commit fa871e211f
117 changed files with 15062 additions and 3874 deletions
@@ -0,0 +1,127 @@
+/*
+Copyright (c) 2022 Advanced Micro Devices, Inc. All rights reserved.
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+The above copyright notice and this permission notice shall be included in
+all copies or substantial portions of the Software.
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+THE SOFTWARE.
+*/
+
+#include "memcpy_performance_common.hh"
+
+/**
+ * @addtogroup memcpy memcpy
+ * @{
+ * @ingroup PerformanceTest
+ */
+
+class Memcpy2DToArrayBenchmark : public Benchmark<Memcpy2DToArrayBenchmark> {
+ public:
+  void operator()(hipArray_t dst, const void* src, size_t src_pitch, size_t width,
+                  size_t height, hipMemcpyKind kind) {
+    TIMED_SECTION(kTimerTypeCpu) {
+      HIP_CHECK(hipMemcpy2DToArray(dst, 0, 0, src, src_pitch, width, height, kind));
+    }
+  }
+};
+
+static void RunBenchmark(size_t width, size_t height, hipMemcpyKind kind,
+                         bool enable_peer_access=false) {
+  Memcpy2DToArrayBenchmark benchmark;
+  benchmark.AddSectionName("(" + std::to_string(width) + ", " + std::to_string(height) + ")");
+
+  if (kind == hipMemcpyHostToDevice) {
+    size_t allocation_size = width * height * sizeof(int);
+    LinearAllocGuard<int> host_allocation(LinearAllocs::hipHostMalloc, allocation_size);
+    ArrayAllocGuard<int> array_allocation(make_hipExtent(width, height, 0), hipArrayDefault);
+    benchmark.Run(array_allocation.ptr(), host_allocation.ptr(), width * sizeof(int),
+                  width * sizeof(int), height, hipMemcpyHostToDevice);
+  } else {
+    // hipMemcpyDeviceToDevice
+    int src_device = std::get<0>(GetDeviceIds(enable_peer_access));
+    int dst_device = std::get<1>(GetDeviceIds(enable_peer_access));
+
+    LinearAllocGuard2D<int> device_allocation(width, height);
+    HIP_CHECK(hipSetDevice(dst_device));
+    ArrayAllocGuard<int> array_allocation(make_hipExtent(width, height, 0), hipArrayDefault);
+    HIP_CHECK(hipSetDevice(src_device));
+    benchmark.Run(array_allocation.ptr(), device_allocation.ptr(), device_allocation.pitch(),
+                  device_allocation.width(), device_allocation.height(), hipMemcpyDeviceToDevice);
+  }
+}
+
+/**
+ * Test Description
+ * ------------------------
+ *  - Executes `hipMemcpy2DToArray` from Host to Device:
+ *    -# Allocation size
+ *      - Small: 4 KB x 32 B
+ *      - Medium: 8 KB x 32 B
+ *      - Large: 16 KB x 32 B
+ * Test source
+ * ------------------------
+ *  - performance/memcpy/hipMemcpy2DToArray.cc
+ * Test requirements
+ * ------------------------
+ *  - HIP_VERSION >= 5.2
+ */
+TEST_CASE("Performance_hipMemcpy2DToArray_HostToDevice") {
+  const auto width = GENERATE(4_KB, 8_KB, 16_KB);
+  RunBenchmark(width, 32, hipMemcpyHostToDevice);
+}
+
+/**
+ * Test Description
+ * ------------------------
+ *  - Executes `hipMemcpy2DToArray` from Device to Device with peer access disabled:
+ *    -# Allocation size
+ *      - Small: 4 KB x 32 B
+ *      - Medium: 8 KB x 32 B
+ *      - Large: 16 KB x 32 B
+ * Test source
+ * ------------------------
+ *  - performance/memcpy/hipMemcpy2DToArray.cc
+ * Test requirements
+ * ------------------------
+ *  - HIP_VERSION >= 5.2
+ */
+TEST_CASE("Performance_hipMemcpy2DToArray_DeviceToDevice_DisablePeerAccess") {
+  const auto width = GENERATE(4_KB, 8_KB, 16_KB);
+  RunBenchmark(width, 32, hipMemcpyDeviceToDevice);
+}
+
+/**
+ * Test Description
+ * ------------------------
+ *  - Executes `hipMemcpy2DToArray` from Device to Device with peer access enabled:
+ *    -# Allocation size
+ *      - Small: 4 KB x 32 B
+ *      - Medium: 8 KB x 32 B
+ *      - Large: 16 KB x 32 B
+ * Test source
+ * ------------------------
+ *  - performance/memcpy/hipMemcpy2DToArray.cc
+ * Test requirements
+ * ------------------------
+ *  - Multi-device
+ *  - Device supports Peer-to-Peer access
+ *  - HIP_VERSION >= 5.2
+ */
+TEST_CASE("Performance_hipMemcpy2DToArray_DeviceToDevice_EnablePeerAccess") {
+  if (HipTest::getDeviceCount() < 2) {
+    HipTest::HIP_SKIP_TEST("This test requires 2 GPUs. Skipping.");
+    return;
+  }
+  const auto width = GENERATE(4_KB, 8_KB, 16_KB);
+  RunBenchmark(width, 32, hipMemcpyDeviceToDevice, true);
+}