Merge branch 'main' into amyachev/device

Author: anmyachev

.github/ISSUE_TEMPLATE/feature-request.yml

@@ -0,0 +1,15 @@
+name: 🚀 Feature request
+description: Submit a request for a new Intel XPU Backend for Triton feature
+
+body:
+- type: textarea
+  attributes:
+    label: 🚀 Feature description
+    description: >
+      A clear and concise description of the feature request.
+    validations:
+      required: true
+- type: markdown
+  attributes:
+    value: >
+      Thanks for contributing 🎉

.github/workflows/build-test-windows.yml

@@ -46,7 +46,7 @@ jobs:
         run: |
           cd ${{ env.NEW_WORKSPACE }}
           cd python
-          pip install -U wheel pybind11 certifi cython cmake
+          pip install -U wheel pybind11 certifi cython cmake setuptools>=65.6.1
           python -m certifi
           pip install -v --no-build-isolation '.[build]'
 

.github/workflows/llvm-build.yml

@@ -30,10 +30,9 @@ jobs:
         - {runner: 'Ubuntu 20.04 ARM64', runs_on: 'ubuntu-20.04', target-os: 'ubuntu', arch: 'arm64'}
         - {runner: 'CentOS 7', runs_on: ['self-hosted', 'CPU'], target-os: 'centos', arch: 'x64'}
         - {runner: 'AlmaLinux 8', runs_on: ['self-hosted', 'CPU'], target-os: 'almalinux', arch: 'x64'}
-        - {runner: 'MacOS X64', runs_on: 'macos-12', target-os: 'macos', arch: 'x64'}
-        - {runner: 'MacOS ARM64', runs_on: 'macos-12', target-os: 'macos', arch: 'arm64'}
-        # TODO(#2805): add back once the workflow works and runs in comparable time to the other ones
-        # - {runner: 'Windows Latest', runs_on: 'windows-latest', target-os: 'windows', arch: 'x64'}
+        - {runner: 'MacOS X64', runs_on: 'macos-13', target-os: 'macos', arch: 'x64'}
+        - {runner: 'MacOS ARM64', runs_on: 'macos-13', target-os: 'macos', arch: 'arm64'}
+        - {runner: 'Windows Latest', runs_on: 'windows-latest', target-os: 'windows', arch: 'x64'}
 
     steps:
 
@@ -127,7 +126,8 @@ jobs:
         -DLLVM_BUILD_TOOLS=ON
         -DLLVM_ENABLE_ASSERTIONS=ON
         -DMLIR_ENABLE_BINDINGS_PYTHON=ON
-        -DLLVM_ENABLE_PROJECTS="clang;mlir"
+        -DLLVM_ENABLE_PROJECTS="mlir;llvm"
+        -DLLVM_ENABLE_DIA_SDK=OFF
         -DLLVM_INSTALL_UTILS=ON
         -DLLVM_TARGETS_TO_BUILD="host;NVPTX;AMDGPU"
         -DLLVM_ENABLE_TERMINFO=OFF
@@ -300,6 +300,7 @@ jobs:
 
     - name: Upload LLVM Artifacts to Azure
       if: ${{ (github.repository == 'triton-lang/triton') }}
+      shell: bash -el {0}
       run: |
         az storage blob upload --account-name oaitriton --auth-mode login --container-name public --file "${{ env.llvm_install_dir }}.tar.gz" --name "llvm-builds/${{ env.llvm_install_dir }}.tar.gz" --overwrite
 

.github/workflows/triton-benchmarks.yml

@@ -250,7 +250,6 @@ jobs:
         run: |
           cd benchmarks/triton_kernels_benchmark
           TRITON_INTEL_ADVANCED_PATH=1 \
-          TRITON_INTEL_ENABLE_ADDRESS_PAYLOAD_OPT=1 \
           IGC_VISAOptions=" -enableBCR" \
           python flash_attention_fwd_benchmark.py --reports $REPORTS
 

.pre-commit-config.yaml

@@ -61,14 +61,6 @@ repos:
       files: ^.github/workflows/integration-tests.yml.*
       pass_filenames: false
 
-  - repo: https://github.com/PyCQA/bandit
-    rev: '1.7.9'
-    hooks:
-    - id: bandit
-      files: '^(benchmarks|scripts|third_party/intel)/.*\.py$'
-      args: ["-c", "bandit.yaml", "-s", "B404,B603,B607"]
-      stages: [pre-commit, pre-push, manual]
-
   - repo: https://github.com/pycqa/pylint
     rev: v3.2.6
     hooks:

CMakeLists.txt

@@ -70,7 +70,6 @@ else()
   set(CMAKE_EXE_LINKER_FLAGS_TRITONRELBUILDWITHASSERTS "/debug:fastlink /INCREMENTAL")
   set(CMAKE_MODULE_LINKER_FLAGS_TRITONRELBUILDWITHASSERTS "/debug:fastlink /INCREMENTAL")
   set(CMAKE_SHARED_LINKER_FLAGS_TRITONRELBUILDWITHASSERTS "/debug:fastlink /INCREMENTAL")
-  set(CMAKE_STATIC_LINKER_FLAGS_TRITONRELBUILDWITHASSERTS "/debug:fastlink /INCREMENTAL")
 endif()
 
 # Default build type
@@ -153,7 +152,7 @@ endfunction()
 if(NOT MSVC)
   set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Werror -Wno-covered-switch-default -fvisibility=hidden")
 else()
-  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /WX- /wd4244 /wd4624 /wd4715 /wd4530")
+  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /wd4244 /wd4624 /wd4715 /wd4530")
 endif()
 
 include_directories(".")

benchmarks/triton_kernels_benchmark/benchmark_driver.py

@@ -133,7 +133,7 @@ def format_of(ty):
             "int8_t": "b",
             "int16_t": "h",
             "int32_t": "i",
-            "int64_t": "l",
+            "int64_t": "L",
             "uint8_t": "B",
             "uint16_t": "H",
             "uint32_t": "I",

bin/CMakeLists.txt

@@ -1,6 +1,5 @@
 get_property(dialect_libs GLOBAL PROPERTY MLIR_DIALECT_LIBS)
 get_property(conversion_libs GLOBAL PROPERTY MLIR_CONVERSION_LIBS)
-get_property(translation_libs GLOBAL PROPERTY MLIR_TRANSLATION_LIBS)
 get_property(triton_libs GLOBAL PROPERTY TRITON_LIBS)
 
 add_llvm_executable(triton-opt triton-opt.cpp PARTIAL_SOURCES_INTENDED)
@@ -104,25 +103,3 @@ target_link_libraries(triton-tensor-layout PRIVATE
   ${dialect_libs}
   TritonTestAnalysis
   )
-
-add_llvm_executable(triton-translate
-  triton-translate.cpp
-
-  PARTIAL_SOURCES_INTENDED
-  DEPENDS
-  intrinsics_gen
-  SUPPORT_PLUGINS
-)
-llvm_update_compile_flags(triton-translate)
-target_link_libraries(triton-translate
-  PRIVATE
-  ${dialect_libs}
-  ${translation_libs}
-  MLIRIR
-  MLIRParser
-  MLIRPass
-  MLIRTranslateLib
-  MLIRSupport
-  TritonGENToLLVMIRTranslation
-)
-mlir_check_link_libraries(triton-translate)

bin/triton-llvm-opt.cpp

@@ -1,7 +1,7 @@
 /// Trimmed down clone of llvm opt to be able to test triton custom llvm ir
 /// passes.
 #include "lib/Target/LLVMIR/LLVMPasses.h"
-#include "third_party/intel/lib/LLVMIR/LLVMPasses.h"
+#include "third_party/intel/lib/Target/LLVMIR/LLVMPasses.h"
 #include "llvm/CodeGen/CommandFlags.h"
 #include "llvm/IR/Constants.h"
 #include "llvm/IR/DataLayout.h"

cmake/AddTritonUnitTest.cmake

@@ -28,7 +28,9 @@ function(add_triton_ut)
           gmock
           ${__LIBS})
 
-  target_compile_options(${__NAME} PRIVATE -fno-rtti)
+  if(NOT MSVC)
+    target_compile_options(${__NAME} PRIVATE -fno-rtti)
+  endif()
 
   target_compile_definitions(${__NAME} PRIVATE ${__DEFS})