[ROCM] Enable ROCM Backend #1: Empty Kernel (triton-lang#1312)

This PR is a first in a series of PRs to import the changes that we have
made to enable ROCM on [our
fork](https://github.com/ROCmSoftwarePlatform/triton) of triton.

The PR contains the major changes to the python frontend and enough
changes to the c++ backend to allow compilation and running of the empty
kernel. We use the ROCM ci added a few weeks ago to verify things.

---------

Co-authored-by: Ronan Keryell <ronan@keryell.fr>

Author: micmelesse

.github/workflows/integration-tests.yml

@@ -22,7 +22,7 @@ jobs:
         id: set-matrix
         run: |
           if [ x"${{ github.repository }}" == x"openai/triton" ]; then
-            echo '::set-output name=matrix::[["self-hosted", "A100"], ["self-hosted", "V100"], "macos-10.15"]'
+            echo '::set-output name=matrix::[["self-hosted", "A100"], ["self-hosted", "V100"], ["self-hosted", "gfx908"], "macos-10.15"]'
           else
             echo '::set-output name=matrix::["ubuntu-latest", "macos-10.15"]'
           fi
@@ -40,6 +40,16 @@ jobs:
       - name: Checkout
         uses: actions/checkout@v2
 
+      - name: Set CUDA ENV
+        if: ${{(matrix.runner[0] == 'self-hosted') && (matrix.runner[1] == 'V100' || matrix.runner[1] == 'A100')}}
+        run: |
+          echo "BACKEND=CUDA" >> $GITHUB_ENV
+
+      - name: Set ROCM ENV
+        if: ${{(matrix.runner[0] == 'self-hosted') && (matrix.runner[1] == 'gfx908')}}
+        run: |
+          echo "BACKEND=ROCM" >> $GITHUB_ENV
+
       - name: Clear cache
         run: |
           rm -rf ~/.triton/
@@ -74,12 +84,22 @@ jobs:
           flake8 --config ./python/setup.cfg ./python || ( echo '::error::Flake8 failed; see logs for errors.' ; exit 1 )
 
       - name: Install Triton
+        if: ${{ env.BACKEND != 'ROCM'}}
         run: |
           cd python
           pip3 install cmake==3.24
           TRITON_USE_ASSERT_ENABLED_LLVM=TRUE pip3 install -e '.[tests]'
 
+      - name: Install Triton on ROCM
+        if: ${{ env.BACKEND == 'ROCM'}}
+        run: |
+          cd python
+          pip3 uninstall --yes torch torchvision torchaudio
+          pip3 install --no-cache-dir --force-reinstall torch==1.13.1 --extra-index-url https://download.pytorch.org/whl/rocm5.2
+          TRITON_USE_ASSERT_ENABLED_LLVM=TRUE pip3 install -e '.[tests]'
+
       - name: Run lit tests
+        if: ${{ env.BACKEND != 'ROCM'}}
         run: |
           pip3 install lit
           cd python
@@ -89,13 +109,20 @@ jobs:
           fi
           lit -v "$LIT_TEST_DIR"
 
-      - name: Run python tests
-        if: ${{matrix.runner[0] == 'self-hosted'}}
+      - name: Run python tests on CUDA
+        if: ${{ env.BACKEND == 'CUDA'}}
         run: |
           cd python/test/unit/
           pytest
+      
+      - name: Run python tests on ROCM
+        if: ${{ env.BACKEND == 'ROCM'}}
+        run: |
+          cd python/test/unit/language/
+          pytest --capture=tee-sys -rfs --verbose "test_core.py::test_empty_kernel"
 
       - name: Run CXX unittests
+        if: ${{ env.BACKEND != 'ROCM'}}
         run: |
           cd python/
           cd "build/$(ls build)"

.gitignore

@@ -216,6 +216,7 @@ if(TRITON_BUILD_PYTHON_MODULE)
     TritonGPUTransforms
     TritonLLVMIR
     TritonPTX
+    TritonHSACO
     ${dialect_libs}
     ${conversion_libs}
 
@@ -228,6 +229,7 @@ if(TRITON_BUILD_PYTHON_MODULE)
     MLIRExecutionEngine
     MLIRMathToLLVM
     MLIRNVVMToLLVMIRTranslation
+    MLIRROCDLToLLVMIRTranslation
     MLIRIR
   )
 

CMakeLists.txt

@@ -30,6 +30,7 @@ llvm_update_compile_flags(triton-translate)
          TritonGPUTransforms
          TritonLLVMIR
          TritonPTX
+         TritonHSACO
          ${dialect_libs}
          ${conversion_libs}
          # tests
@@ -53,5 +54,6 @@ llvm_update_compile_flags(triton-translate)
          MLIRTransformUtils
          MLIRLLVMToLLVMIRTranslation
          MLIRNVVMToLLVMIRTranslation
+         MLIRROCDLToLLVMIRTranslation
          )
 mlir_check_all_link_libraries(triton-translate)

bin/CMakeLists.txt

@@ -4,7 +4,8 @@
 #include "triton/Dialect/Triton/Transforms/Passes.h"
 #include "triton/Dialect/TritonGPU/Transforms/Passes.h"
 
-#include "triton/Conversion/Passes.h"
+#include "triton/Conversion/TritonGPUToLLVM/Passes.h"
+#include "triton/Conversion/TritonToTritonGPU/Passes.h"
 
 #include "mlir/IR/Dialect.h"
 #include "mlir/InitAllPasses.h"

bin/triton-opt.cpp

@@ -14,14 +14,14 @@
 #include "triton/Conversion/TritonToTritonGPU/TritonToTritonGPUPass.h"
 #include "triton/Dialect/Triton/IR/Dialect.h"
 #include "triton/Dialect/TritonGPU/IR/Dialect.h"
+#include "triton/Target/HSACO/HSACOTranslation.h"
 #include "triton/Target/LLVMIR/LLVMIRTranslation.h"
 #include "triton/Target/PTX/PTXTranslation.h"
 #include "llvm/IR/LLVMContext.h"
 #include "llvm/Support/CommandLine.h"
 #include "llvm/Support/InitLLVM.h"
 #include "llvm/Support/SourceMgr.h"
 #include "llvm/Support/ToolOutputFile.h"
-#include <iostream>
 
 namespace mlir {
 namespace triton {
@@ -79,7 +79,8 @@ LogicalResult tritonTranslateMain(int argc, char **argv,
       llvm::cl::init("-"));
 
   static llvm::cl::opt<std::string> targetKind(
-      "target", llvm::cl::desc("<translation target, options: llvmir/ptx>"),
+      "target",
+      llvm::cl::desc("<translation target, options: llvmir/ptx/hsaco>"),
       llvm::cl::value_desc("target"), llvm::cl::init("llvmir"));
 
   static llvm::cl::opt<int> SMArch("sm", llvm::cl::desc("sm arch"),
@@ -88,6 +89,18 @@ LogicalResult tritonTranslateMain(int argc, char **argv,
   static llvm::cl::opt<int> ptxVersion(
       "ptx-version", llvm::cl::desc("PTX version"), llvm::cl::init(10000));
 
+  static llvm::cl::opt<std::string> GCNArch(
+      "gfx", llvm::cl::desc("AMDGCN target. e.g. '90a'"),
+      llvm::cl::value_desc("architecture"), llvm::cl::init("90a"));
+
+  static llvm::cl::opt<std::string> GCNTriple(
+      "amdgcn", llvm::cl::desc("AMDGCN triple. e.g. '-amd-amdhsa'"),
+      llvm::cl::value_desc("target triple"), llvm::cl::init("-amd-amdhsa"));
+
+  static llvm::cl::opt<std::string> GCNFeatures(
+      "", llvm::cl::desc("AMDGCN features. e.g. '+sramecc,-xnack'"),
+      llvm::cl::value_desc("features"), llvm::cl::init("+sramecc,-xnack"));
+
   llvm::InitLLVM y(argc, argv);
 
   registerAsmPrinterCLOptions();
@@ -119,6 +132,15 @@ LogicalResult tritonTranslateMain(int argc, char **argv,
   else if (targetKind == "ptx")
     llvm::outs() << ::triton::translateLLVMIRToPTX(*llvmir, SMArch.getValue(),
                                                    ptxVersion.getValue());
+  else if (targetKind == "hsaco") {
+    auto [module, hsaco] = ::triton::translateLLVMIRToHSACO(
+        *llvmir, GCNArch.getValue(), GCNTriple.getValue(),
+        GCNFeatures.getValue());
+    llvm::outs() << hsaco;
+  } else {
+    llvm::errs() << "Error: Unknown target specified: " << targetKind << "\n";
+    return failure();
+  }
 
   return success();
 }

bin/triton-translate.cpp

@@ -1,3 +1,2 @@
-set(LLVM_TARGET_DEFINITIONS Passes.td)
-mlir_tablegen(Passes.h.inc -gen-pass-decls)
-add_public_tablegen_target(TritonConversionPassIncGen)
+add_subdirectory(TritonToTritonGPU)
+add_subdirectory(TritonGPUToLLVM)

include/triton/Conversion/CMakeLists.txt

@@ -0,0 +1,33 @@
+#ifndef TRITON_CONVERSION_TRITON_GPU_TO_LLVM_ASM_FORMAT_H_
+#define TRITON_CONVERSION_TRITON_GPU_TO_LLVM_ASM_FORMAT_H_
+
+#include "mlir/IR/Value.h"
+#include "triton/Dialect/Triton/IR/Dialect.h"
+#include "llvm/ADT/SmallVector.h"
+#include "llvm/ADT/StringRef.h"
+#include <memory>
+#include <string>
+
+namespace mlir {
+class ConversionPatternRewriter;
+class Location;
+
+namespace triton {
+using llvm::StringRef;
+
+inline std::string strJoin(llvm::ArrayRef<std::string> strs,
+                           llvm::StringRef delimiter) {
+  std::string osStr;
+  llvm::raw_string_ostream os(osStr);
+  for (size_t i = 0; !strs.empty() && i < strs.size() - 1; ++i)
+    os << strs[i] << delimiter;
+  if (!strs.empty())
+    os << strs.back();
+  os.flush();
+  return osStr;
+}
+
+} // namespace triton
+} // namespace mlir
+
+#endif // TRITON_CONVERSION_TRITON_GPU_TO_LLVM_ASM_FORMAT_H_

include/triton/Conversion/TritonGPUToLLVM/AsmFormat.h

@@ -0,0 +1,3 @@
+set(LLVM_TARGET_DEFINITIONS Passes.td)
+mlir_tablegen(Passes.h.inc -gen-pass-decls --name TritonGPUToLLVM)
+add_public_tablegen_target(TritonGPUConversionPassIncGen)