SWDEV-1 - Merge github PRs to amd-staging

- #194 - #36 - #44 - #47 - #62 - #63 - #64 - #65 - #66 - #67 - #68 - #69 - #142 - #196 - #238 Change-Id: I74f7fef76d7d536b1cf89dad3e527c92d1cd21b5
ROCm · Dec 20, 2023 · 6429ef1 · 6429ef1
2 parents 6d8e62c + e0612f9
commit 6429ef1
Show file tree

Hide file tree

Showing 99 changed files with 9,848 additions and 2,565 deletions.
diff --git a/catch/hipTestMain/config/config_amd_linux b/catch/hipTestMain/config/config_amd_linux
@@ -54,6 +54,8 @@
         "Unit_hipFuncSetAttribute_Positive_MaxDynamicSharedMemorySize_Not_Supported",
         "Unit_hipFuncSetAttribute_Positive_PreferredSharedMemoryCarveout_Not_Supported",
         "Unit_hipOccupancyMaxActiveBlocksPerMultiprocessor_Negative_Parameters",
+        "Unit_hipModuleOccupancyMaxActiveBlocksPerMultiprocessorWithFlags_Negative_Parameters",
+        "Unit_hipModuleOccupancyMaxPotentialBlockSizeWithFlags_Negative_Parameters",
         "Unit_hipGraphMemcpyNodeSetParamsToSymbol_Positive_Basic",
         "Unit_hipGraphExecMemcpyNodeSetParamsToSymbol_Positive_Basic",
         "Unit_hipGraphExecMemcpyNodeSetParamsFromSymbol_Positive_Basic",
@@ -183,6 +185,62 @@
         "Unit_hipMemUnmap_negative",
         "=== SWDEV-432556,SWDEV-434211:Below test randomly failing in stress test ===",
         "Unit_hipDeviceGetUuid_From_RocmInfo",
+        "=== SWDEV-434171: Below tests took long time to complete in stress test on 17/11/23 ===",
+        "Unit_Warp_Shfl_Positive_Basic - int",
+        "Unit_Warp_Shfl_Positive_Basic - unsigned int",
+        "Unit_Warp_Shfl_Positive_Basic - long",
+        "Unit_Warp_Shfl_Positive_Basic - unsigned long",
+        "Unit_Warp_Shfl_Positive_Basic - long long",
+        "Unit_Warp_Shfl_Positive_Basic - unsigned long long",
+        "Unit_Warp_Shfl_Positive_Basic - float",
+        "Unit_Warp_Shfl_Positive_Basic - double",
+        "Unit_Warp_Shfl_XOR_Positive_Basic - int",
+        "Unit_Warp_Shfl_XOR_Positive_Basic - unsigned int",
+        "Unit_Warp_Shfl_XOR_Positive_Basic - long",
+        "Unit_Warp_Shfl_XOR_Positive_Basic - unsigned long",
+        "Unit_Warp_Shfl_XOR_Positive_Basic - long long",
+        "Unit_Warp_Shfl_XOR_Positive_Basic - unsigned long long",
+        "Unit_Warp_Shfl_XOR_Positive_Basic - float",
+        "Unit_Warp_Shfl_XOR_Positive_Basic - double",
+        "=== SWDEV-434878: Below tests failed in stress test on 24/11/23 ===",
+        "Unit_hipGraphUpload_Negative_Parameters",
+        "Unit_hipModuleOccupancyMaxPotentialBlockSize_Negative_Parameters",
+        "Unit_hipModuleOccupancyMaxPotentialBlockSize_Positive_RangeValidation",
+        "Unit_hipModuleOccupancyMaxPotentialBlockSizeWithFlags_Positive_RangeValidation",
+        "Unit_hipModuleOccupancyMaxActiveBlocksPerMultiprocessor_Negative_Parameters",
+        "Unit_hipModuleOccupancyMaxActiveBlocksPerMultiprocessor_Positive_RangeValidation",
+        "Unit_hipModuleOccupancyMaxActiveBlocksPerMultiprocessorWithFlags_Positive_RangeValidation",
+        "=== SWDEV-435667: Below tests failing randomly in stress test on 01/12/23 ===",
+        "Unit_atomicExch_Positive_Same_Address_Compile_Time - int",
+        "Unit_atomicExch_Positive_Same_Address_Compile_Time - unsigned int",
+        "Unit_atomicExch_Positive_Same_Address_Compile_Time - unsigned long",
+        "Unit_atomicExch_Positive_Same_Address_Compile_Time - unsigned long long",
+        "Unit_atomicExch_Positive_Same_Address_Compile_Time - float",
+        "Unit_atomicExch_Positive_Same_Address_Compile_Time - double",
+        "Unit_atomicExch_Positive_Multi_Kernel - int",
+        "Unit_atomicExch_Positive_Multi_Kernel - unsigned int",
+        "Unit_atomicExch_Positive_Multi_Kernel - unsigned long",
+        "Unit_atomicExch_Positive_Multi_Kernel - unsigned long long",
+        "Unit_atomicExch_Positive_Multi_Kernel - float",
+        "Unit_atomicExch_Positive_Multi_Kernel - double",
+        "Unit_atomicExch_system_Positive_Peer_GPUs - int",
+        "Unit_atomicExch_system_Positive_Peer_GPUs - unsigned int",
+        "Unit_atomicExch_system_Positive_Peer_GPUs - unsigned long",
+        "Unit_atomicExch_system_Positive_Peer_GPUs - unsigned long long",
+        "Unit_atomicExch_system_Positive_Peer_GPUs - float",
+        "Unit_atomicExch_system_Positive_Peer_GPUs - double",
+        "Unit_atomicExch_system_Positive_Host_And_GPU - int",
+        "Unit_atomicExch_system_Positive_Host_And_GPU - unsigned int",
+        "Unit_atomicExch_system_Positive_Host_And_GPU - unsigned long",
+        "Unit_atomicExch_system_Positive_Host_And_GPU - unsigned long long",
+        "Unit_atomicExch_system_Positive_Host_And_GPU - float",
+        "Unit_atomicExch_system_Positive_Host_And_GPU - double",
+        "Unit_atomicExch_system_Positive_Host_And_Peer_GPUs - int",
+        "Unit_atomicExch_system_Positive_Host_And_Peer_GPUs - unsigned int",
+        "Unit_atomicExch_system_Positive_Host_And_Peer_GPUs - unsigned long",
+        "Unit_atomicExch_system_Positive_Host_And_Peer_GPUs - unsigned long long",
+        "Unit_atomicExch_system_Positive_Host_And_Peer_GPUs - float",
+        "Unit_atomicExch_system_Positive_Host_And_Peer_GPUs - double",
     #endif
     #if defined VEGA20
         "=== SWDEV-419112 Below tests fail in stress test on 29/08/23 ===",

diff --git a/catch/hipTestMain/config/config_amd_windows b/catch/hipTestMain/config/config_amd_windows
@@ -119,6 +119,8 @@
         "Unit_hipFuncSetAttribute_Positive_MaxDynamicSharedMemorySize_Not_Supported",
         "Unit_hipFuncSetAttribute_Positive_PreferredSharedMemoryCarveout_Not_Supported",
         "Unit_hipOccupancyMaxActiveBlocksPerMultiprocessor_Negative_Parameters",
+        "Unit_hipModuleOccupancyMaxActiveBlocksPerMultiprocessorWithFlags_Negative_Parameters",
+        "Unit_hipModuleOccupancyMaxPotentialBlockSizeWithFlags_Negative_Parameters",
         "Unit_hipGraphMemcpyNodeSetParamsToSymbol_Positive_Basic",
         "Unit_hipGraphExecMemcpyNodeSetParamsToSymbol_Positive_Basic",
         "Unit_hipGraphMemcpyNodeSetParamsFromSymbol_Positive_Basic",
@@ -282,6 +284,62 @@
         "Unit_hipMemSetAccess_MultiProc",
         "Unit_hipMemSetAccess_negative",
         "Unit_hipMemUnmap_negative",
+        "=== SWDEV-434171: Below tests took long time to complete in stress test on 17/11/23 ===",
+        "Unit_Warp_Shfl_Positive_Basic - int",
+        "Unit_Warp_Shfl_Positive_Basic - unsigned int",
+        "Unit_Warp_Shfl_Positive_Basic - long",
+        "Unit_Warp_Shfl_Positive_Basic - unsigned long",
+        "Unit_Warp_Shfl_Positive_Basic - long long",
+        "Unit_Warp_Shfl_Positive_Basic - unsigned long long",
+        "Unit_Warp_Shfl_Positive_Basic - float",
+        "Unit_Warp_Shfl_Positive_Basic - double",
+        "Unit_Warp_Shfl_XOR_Positive_Basic - int",
+        "Unit_Warp_Shfl_XOR_Positive_Basic - unsigned int",
+        "Unit_Warp_Shfl_XOR_Positive_Basic - long",
+        "Unit_Warp_Shfl_XOR_Positive_Basic - unsigned long",
+        "Unit_Warp_Shfl_XOR_Positive_Basic - long long",
+        "Unit_Warp_Shfl_XOR_Positive_Basic - unsigned long long",
+        "Unit_Warp_Shfl_XOR_Positive_Basic - float",
+        "Unit_Warp_Shfl_XOR_Positive_Basic - double",
+        "=== SWDEV-434878: Below tests failed in stress test on 24/11/23 ===",
+        "Unit_hipGraphUpload_Negative_Parameters",
+        "Unit_hipModuleOccupancyMaxPotentialBlockSize_Negative_Parameters",
+        "Unit_hipModuleOccupancyMaxPotentialBlockSize_Positive_RangeValidation",
+        "Unit_hipModuleOccupancyMaxPotentialBlockSizeWithFlags_Positive_RangeValidation",
+        "Unit_hipModuleOccupancyMaxActiveBlocksPerMultiprocessor_Negative_Parameters",
+        "Unit_hipModuleOccupancyMaxActiveBlocksPerMultiprocessor_Positive_RangeValidation",
+        "Unit_hipModuleOccupancyMaxActiveBlocksPerMultiprocessorWithFlags_Positive_RangeValidation",
+        "=== SWDEV-435667: Below tests failing randomly in stress test on 01/12/23 ===",
+        "Unit_atomicExch_Positive_Same_Address_Compile_Time - int",
+        "Unit_atomicExch_Positive_Same_Address_Compile_Time - unsigned int",
+        "Unit_atomicExch_Positive_Same_Address_Compile_Time - unsigned long",
+        "Unit_atomicExch_Positive_Same_Address_Compile_Time - unsigned long long",
+        "Unit_atomicExch_Positive_Same_Address_Compile_Time - float",
+        "Unit_atomicExch_Positive_Same_Address_Compile_Time - double",
+        "Unit_atomicExch_Positive_Multi_Kernel - int",
+        "Unit_atomicExch_Positive_Multi_Kernel - unsigned int",
+        "Unit_atomicExch_Positive_Multi_Kernel - unsigned long",
+        "Unit_atomicExch_Positive_Multi_Kernel - unsigned long long",
+        "Unit_atomicExch_Positive_Multi_Kernel - float",
+        "Unit_atomicExch_Positive_Multi_Kernel - double",
+        "Unit_atomicExch_system_Positive_Peer_GPUs - int",
+        "Unit_atomicExch_system_Positive_Peer_GPUs - unsigned int",
+        "Unit_atomicExch_system_Positive_Peer_GPUs - unsigned long",
+        "Unit_atomicExch_system_Positive_Peer_GPUs - unsigned long long",
+        "Unit_atomicExch_system_Positive_Peer_GPUs - float",
+        "Unit_atomicExch_system_Positive_Peer_GPUs - double",
+        "Unit_atomicExch_system_Positive_Host_And_GPU - int",
+        "Unit_atomicExch_system_Positive_Host_And_GPU - unsigned int",
+        "Unit_atomicExch_system_Positive_Host_And_GPU - unsigned long",
+        "Unit_atomicExch_system_Positive_Host_And_GPU - unsigned long long",
+        "Unit_atomicExch_system_Positive_Host_And_GPU - float",
+        "Unit_atomicExch_system_Positive_Host_And_GPU - double",
+        "Unit_atomicExch_system_Positive_Host_And_Peer_GPUs - int",
+        "Unit_atomicExch_system_Positive_Host_And_Peer_GPUs - unsigned int",
+        "Unit_atomicExch_system_Positive_Host_And_Peer_GPUs - unsigned long",
+        "Unit_atomicExch_system_Positive_Host_And_Peer_GPUs - unsigned long long",
+        "Unit_atomicExch_system_Positive_Host_And_Peer_GPUs - float",
+        "Unit_atomicExch_system_Positive_Host_And_Peer_GPUs - double",
     #endif
         "End of json"
     ]

diff --git a/catch/hipTestMain/config/config_nvidia_linux.json b/catch/hipTestMain/config/config_nvidia_linux.json
@@ -44,6 +44,14 @@
         "Grid_Group_Getters_Via_Non_Member_Functions_Positive_Basic",
         "Grid_Group_Sync_Positive_Basic",
         "dynamic_loading_device_kernels_from_library",
-        "Unit_tiled_partition"
+        "Unit_tiled_partition",
+        "Unit_atomicExch_Positive_Same_Address_Compile_Time - int",
+        "Unit_atomicExch_Positive_Same_Address_Compile_Time - unsigned int",
+        "Unit_atomicExch_Positive_Same_Address_Compile_Time - unsigned long long",
+        "Unit_atomicExch_Positive_Same_Address_Compile_Time - float",
+        "Unit_atomicExch_system_Positive_Host_And_GPU - int",
+        "Unit_atomicExch_system_Positive_Host_And_GPU - unsigned int",
+        "Unit_atomicExch_system_Positive_Host_And_GPU - unsigned long long",
+        "Unit_atomicExch_system_Positive_Host_And_GPU - float"
     ]
 }
diff --git a/catch/hipTestMain/main.cc b/catch/hipTestMain/main.cc
@@ -30,9 +30,9 @@ int main(int argc, char** argv) {
     | Opt(cmd_options.progress)
         ["-P"]["--progress"]
         ("Show progress bar when running performance tests")
-    | Opt(cmd_options.extended_run)
-        ["-E"]["--extended-run"]
-        ("TODO: Description goes here")
+    | Opt(cmd_options.cg_extended_run, "cg_extened_run")
+        ["-E"]["--cg-extended-run"]
+        ("Number of iterations used for cooperative groups sync tests (default: 5)")
   ;
   // clang-format on
 

diff --git a/catch/include/cmd_options.hh b/catch/include/cmd_options.hh
@@ -23,11 +23,11 @@ THE SOFTWARE.
 #pragma once
 
 struct CmdOptions {
-  int iterations = 1000;
+  int iterations = 10;
   int warmups = 100;
+  int cg_extended_run = 5;
   bool no_display = false;
   bool progress = false;
-  bool extended_run = false;
 };
 
 extern CmdOptions cmd_options;
diff --git a/catch/include/cpu_grid.h b/catch/include/cpu_grid.h
@@ -78,6 +78,7 @@ struct CPUGrid {
   unsigned int thread_count_;
 };
 
+/* Generate dimensions for 1D, 2D and 3D blocks of threads */
 inline dim3 GenerateThreadDimensions() {
   hipDeviceProp_t props;
   HIP_CHECK(hipGetDeviceProperties(&props, 0));
@@ -99,6 +100,7 @@ inline dim3 GenerateThreadDimensions() {
       dim3(props.warpSize + 1, 3, 3));
 }
 
+/* Generate dimensions for 1D, 2D and 3D grids of blocks */
 inline dim3 GenerateBlockDimensions() {
   hipDeviceProp_t props;
   HIP_CHECK(hipGetDeviceProperties(&props, 0));
@@ -116,6 +118,7 @@ inline dim3 GenerateBlockDimensions() {
                        dim3(5, 5, 5));
 }
 
+/* Generate dimensions for 1D, 2D and 3D blocks of threads - reduced set */
 inline dim3 GenerateThreadDimensionsForShuffle() {
   hipDeviceProp_t props;
   HIP_CHECK(hipGetDeviceProperties(&props, 0));
@@ -136,6 +139,7 @@ inline dim3 GenerateThreadDimensionsForShuffle() {
       dim3(props.warpSize + 1, 3, 3));
 }
 
+/* Generate dimensions for 1D, 2D and 3D grids of blocks - reduced set */
 inline dim3 GenerateBlockDimensionsForShuffle() {
   hipDeviceProp_t props;
   HIP_CHECK(hipGetDeviceProperties(&props, 0));

diff --git a/catch/include/hip_test_common.hh b/catch/include/hip_test_common.hh
@@ -102,6 +102,19 @@ THE SOFTWARE.
     }                                                                                              \
   }
 
+// Check that an expression, errorExpr, evaluates to the expected error_t, expectedError.
+#define HIPRTC_CHECK_ERROR(errorExpr, expectedError)                                               \
+  {                                                                                                \
+    auto localError = errorExpr;                                                                   \
+    INFO("Matching Errors: "                                                                       \
+         << "\n    Expected Error: " << hiprtcGetErrorString(expectedError)                        \
+         << "\n    Expected Code: " << expectedError << '\n'                                       \
+         << "                  Actual Error:   " << hiprtcGetErrorString(localError)               \
+         << "\n    Actual Code:   " << localError << "\nStr: " << #errorExpr                       \
+         << "\n    In File: " << __FILE__ << "\n    At line: " << __LINE__);                       \
+    REQUIRE(localError == expectedError);                                                          \
+  }
+
 // Although its assert, it will be evaluated at runtime
 #define HIP_ASSERT(x)                                                                              \
   { REQUIRE((x)); }

diff --git a/catch/include/hip_test_defgroups.hh b/catch/include/hip_test_defgroups.hh
@@ -20,6 +20,8 @@ OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 THE SOFTWARE.
 */
 
+#include <hip_test_common.hh>
+
 // Test groups are named based on the group names from hip_api_runtime.h, with adding "Test" suffix
 
 /**
@@ -95,8 +97,46 @@ THE SOFTWARE.
 
 /**
 * @defgroup KernelTest Kernel Functions Management
+* @{
+* This section describes the various kernel functions invocation.
+* @}
+*/
+
+/**
+ * @defgroup AtomicsTest Device Atomics
  * @{
- * This section describes the various kernel functions invocation.
+ * This section describes tests for the Device Atomic APIs.
+ * @}
+ */
+
+ /**
+ * @addtogroup atomicExch atomicExch
+ * @{
+ * @ingroup AtomicsTest
+ */
+
+/**
+ * Test Description
+ * ------------------------
+ *  - Compiles atomicExch with invalid parameters.
+ *  - Compiles the source with specialized Python tool.
+ *    -# Utilizes sub-process to invoke compilation of faulty source.
+ *    -# Performs post-processing of compiler output and counts errors.
+ * Test source
+ * ------------------------
+ *  - unit/atomics/CMakeLists.txt
+ * Test requirements
+ * ------------------------
+ *  - HIP_VERSION >= 5.2
+ */
+TEST_CASE("Unit_atomicExch_Negative_Parameters") {}
+/**
+ * End doxygen group atomicExch.
+ * @}
+ */
+
+/**
+ * End doxygen group AtomicsTest.
  * @}
  */
 
@@ -115,7 +155,14 @@ THE SOFTWARE.
  * @}
  */
 
- /**
+/**
+ * @defgroup PerformanceTest Performance tests
+ * @{
+ * This section describes performance tests for the target API groups and use-cases.
+ * @}
+ */
+
+/**
  * @defgroup ShflTest warp shuffle function Management
  * @{
  * This section describes the warp shuffle types & functions of HIP runtime API.

diff --git a/catch/include/performance_common.hh b/catch/include/performance_common.hh
@@ -34,6 +34,7 @@ THE SOFTWARE.
 #include <resource_guards.hh>
 
 #pragma clang diagnostic ignored "-Wunused-but-set-variable"
+#pragma clang diagnostic ignored "-Wunused-parameter"
 #pragma clang diagnostic ignored "-Wunused-function"
 
 #if defined(_WIN32)