Merge branch 'main' into cooperative_kernel_functions

Signed-off-by: Michael Aziz <[email protected]>

Author: 0x12CC

.github/workflows/cmake.yml

@@ -166,10 +166,10 @@ jobs:
     strategy:
       matrix:
         adapter: [
-          {name: CUDA, triplet: nvptx64-nvidia-cuda, platform: ""},
-          {name: HIP, triplet: amdgcn-amd-amdhsa, platform: ""},
-          {name: L0, triplet: spir64, platform: ""},
-          {name: OPENCL, triplet: spir64, platform: "Intel(R) OpenCL"}
+          {name: CUDA, platform: ""},
+          {name: HIP, platform: ""},
+          {name: L0, platform: ""},
+          {name: OPENCL, platform: "Intel(R) OpenCL"}
         ]
         build_type: [Debug, Release]
         compiler: [{c: gcc, cxx: g++}, {c: clang, cxx: clang++}]
@@ -201,8 +201,7 @@ jobs:
         -DUR_BUILD_ADAPTER_${{matrix.adapter.name}}=ON
         -DUR_DPCXX=${{github.workspace}}/dpcpp_compiler/bin/clang++
         -DUR_SYCL_LIBRARY_DIR=${{github.workspace}}/dpcpp_compiler/lib
-        -DUR_CONFORMANCE_TARGET_TRIPLES=${{matrix.adapter.triplet}}
-        ${{ matrix.adapter.name == 'HIP' && '-DAMD_ARCH=gfx1030' || '' }} 
+        ${{ matrix.adapter.name == 'HIP' && '-DUR_CONFORMANCE_AMD_ARCH=gfx1030' || '' }} 
         ${{ matrix.adapter.name == 'HIP' && '-DUR_HIP_PLATFORM=AMD' || '' }}
 
     - name: Build

CMakeLists.txt

@@ -45,10 +45,13 @@ option(UR_BUILD_ADAPTER_NATIVE_CPU "Build the Native-CPU adapter" OFF)
 option(UR_BUILD_ADAPTER_ALL "Build all currently supported adapters" OFF)
 option(UR_BUILD_EXAMPLE_CODEGEN "Build the codegen example." OFF)
 option(VAL_USE_LIBBACKTRACE_BACKTRACE "enable libbacktrace validation backtrace for linux" OFF)
+option(UR_ENABLE_ASSERTIONS "Enable assertions for all build types" OFF)
 set(UR_DPCXX "" CACHE FILEPATH "Path of the DPC++ compiler executable")
 set(UR_SYCL_LIBRARY_DIR "" CACHE PATH
     "Path of the SYCL runtime library directory")
-option(UR_ENABLE_ASSERTIONS "Enable assertions for all build types" OFF)
+set(UR_CONFORMANCE_TARGET_TRIPLES "" CACHE STRING
+    "List of sycl targets to build CTS device binaries for")
+set(UR_CONFORMANCE_AMD_ARCH "" CACHE STRING "AMD device target ID to build CTS binaries for")
 
 include(Assertions)
 

README.md

@@ -128,6 +128,7 @@ List of options provided by CMake:
 | UR_ENABLE_TRACING | Enable XPTI-based tracing layer | ON/OFF | OFF |
 | UR_ENABLE_SANITIZER | Enable device sanitizer layer | ON/OFF | ON |
 | UR_CONFORMANCE_TARGET_TRIPLES | SYCL triples to build CTS device binaries for | Comma-separated list | spir64 |
+| UR_CONFORMANCE_AMD_ARCH | AMD device target ID to build CTS binaries for | string | `""` |
 | UR_BUILD_ADAPTER_L0     | Build the Level-Zero adapter            | ON/OFF     | OFF     |
 | UR_BUILD_ADAPTER_OPENCL | Build the OpenCL adapter                | ON/OFF     | OFF     |
 | UR_BUILD_ADAPTER_CUDA   | Build the CUDA adapter                  | ON/OFF     | OFF     |

cmake/FindRocmAgentEnumerator.cmake

@@ -0,0 +1,19 @@
+# Copyright (C) 2023 Intel Corporation
+# Part of the Unified-Runtime Project, under the Apache License v2.0 with LLVM Exceptions.
+# See LICENSE.TXT
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+#
+# FindRocmAgentEnumerator.cmake -- module searching for rocm_agent_enumerator.
+#                                  ROCM_AGENT_ENUMERATOR_FOUND is set to true if
+#                                  rocm_agent_enumerator is found.
+#
+
+find_program(ROCM_AGENT_ENUMERATOR NAMES not_rocm_agent_enumerator)
+
+if(ROCM_AGENT_ENUMERATOR)
+    set(ROCM_AGENT_ENUMERATOR_FOUND TRUE)
+endif()
+
+include(FindPackageHandleStandardArgs)
+find_package_handle_standard_args(RocmAgentEnumerator DEFAULT_MSG ROCM_AGENT_ENUMERATOR)

include/ur_api.h

@@ -5103,7 +5103,7 @@ urKernelCreateWithNativeHandle(
 ///////////////////////////////////////////////////////////////////////////////
 /// @brief Query queue info
 typedef enum ur_queue_info_t {
-    UR_QUEUE_INFO_CONTEXT = 0,         ///< [::ur_queue_handle_t] context associated with this queue.
+    UR_QUEUE_INFO_CONTEXT = 0,         ///< [::ur_context_handle_t] context associated with this queue.
     UR_QUEUE_INFO_DEVICE = 1,          ///< [::ur_device_handle_t] device associated with this queue.
     UR_QUEUE_INFO_DEVICE_DEFAULT = 2,  ///< [::ur_queue_handle_t] the current default queue of the underlying
                                        ///< device.