Merge pull request #2 from intel/gh/arthuryuan1987/1/head

SYCL building system

CMakeLists.txt

@@ -0,0 +1,33 @@
+# torch-xpu-ops: XPU implementation for PyTorch ATen
+
+# outputs:
+#
+#  PYTORCH_FOUND_XPU
+#  -- The flag to indicate whether XPU backend stacks are setup successfully or not.
+#
+#  libtorch_xpu_ops
+#  -- Static archive library target
+
+cmake_minimum_required(VERSION 3.13 FATAL_ERROR)
+project(${TORCH_XPU_OPS_PROJ_NAME} VERSION ${CMAKE_PROJECT_VERSION})
+
+set(PYTORCH_FOUND_XPU FALSE)
+
+if(NOT CMAKE_SYSTEM_NAME MATCHES "Linux")
+  message("torch-xpu-ops only supports Linux system so far. We will support more systems in future.")
+  return()
+endif()
+
+set(TORCH_XPU_OPS_ROOT ${PROJECT_SOURCE_DIR})
+list(APPEND CMAKE_MODULE_PATH ${TORCH_XPU_OPS_ROOT}/cmake/Modules)
+
+include(${TORCH_XPU_OPS_ROOT}/cmake/SYCL.cmake)
+include(${TORCH_XPU_OPS_ROOT}/cmake/BuildFlags.cmake)
+
+if(BUILD_TEST)
+  add_subdirectory(${TORCH_XPU_OPS_ROOT}/test/sycl ${CMAKE_BINARY_DIR}/test_sycl)
+endif()
+
+set(PYTORCH_FOUND_XPU TRUE)
+
+message(STATUS "XPU found")

cmake/BuildFlags.cmake

@@ -0,0 +1,41 @@
+# Setup building flags for SYCL device and host codes.
+
+# Support GCC only at the moment.
+if(CMAKE_CXX_COMPILER_ID STREQUAL "GNU")
+  # # -- Host flags (SYCL_CXX_FLAGS)
+  list(APPEND SYCL_HOST_FLAGS -fPIC)
+  list(APPEND SYCL_HOST_FLAGS -std=c++17)
+  # SYCL headers warnings
+  list(APPEND SYCL_HOST_FLAGS -Wno-deprecated-declarations)
+  list(APPEND SYCL_HOST_FLAGS -Wno-attributes)
+
+  if(CMAKE_BUILD_TYPE MATCHES Debug)
+    list(APPEND SYCL_HOST_FLAGS -g)
+    list(APPEND SYCL_HOST_FLAGS -O0)
+  endif(CMAKE_BUILD_TYPE MATCHES Debug)
+
+  # -- Kernel flags (SYCL_KERNEL_OPTIONS)
+  # The fast-math will be enabled by default in SYCL compiler.
+  # Refer to [https://clang.llvm.org/docs/UsersManual.html#cmdoption-fno-fast-math]
+  # 1. We enable below flags here to be warn about NaN and Infinity,
+  # which will be hidden by fast-math by default.
+  # 2. The associative-math in fast-math allows floating point
+  # operations to be reassociated, which will lead to non-deterministic
+  # results compared with CUDA backend.
+  # 3. The approx-func allows certain math function calls (such as log, sqrt, pow, etc)
+  # to be replaced with an approximately equivalent set of instructions or
+  # alternative math function calls, which have great errors.
+  set(SYCL_KERNEL_OPTIONS ${SYCL_KERNEL_OPTIONS} -fno-sycl-unnamed-lambda)
+  set(SYCL_KERNEL_OPTIONS ${SYCL_KERNEL_OPTIONS} -sycl-std=2020)
+  set(SYCL_KERNEL_OPTIONS ${SYCL_KERNEL_OPTIONS} -fhonor-nans)
+  set(SYCL_KERNEL_OPTIONS ${SYCL_KERNEL_OPTIONS} -fhonor-infinities)
+  set(SYCL_KERNEL_OPTIONS ${SYCL_KERNEL_OPTIONS} -fno-associative-math)
+  set(SYCL_KERNEL_OPTIONS ${SYCL_KERNEL_OPTIONS} -fno-approx-func)
+  # TODO: Align with PyTorch and switch to ABI=0 eventually, after
+  # resolving incompatible implementation in SYCL runtime.
+  set(SYCL_KERNEL_OPTIONS ${SYCL_KERNEL_OPTIONS} -D_GLIBCXX_USE_CXX11_ABI=1)
+  set(SYCL_FLAGS ${SYCL_FLAGS} ${SYCL_KERNEL_OPTIONS})
+else()
+  message("Not compiling with XPU. Only support GCC compiler as CXX compiler.")
+  return()
+endif()