[Mlir-commits] [mlir] [MLIR][XeGPU] Updates XeGPU TensorDescAttr and Refine Gather/Scatter definition. (PR #109144)

LLVM Continuous Integration llvmlistbot at llvm.org
Mon Sep 23 07:03:34 PDT 2024 

llvm-ci wrote:

LLVM Buildbot has detected a new failure on builder `mlir-nvidia-gcc7` running on `mlir-nvidia` while building `mlir` at step 5 "build-check-mlir-build-only".

Full details are available at: https://lab.llvm.org/buildbot/#/builders/116/builds/4036

<details>
<summary>Here is the relevant piece of the build log for the reference</summary>

```
Step 5 (build-check-mlir-build-only) failure: build (failure)
...
27.618 [290/16/4255] Linking CXX static library lib/libLLVMFrontendOffloading.a
27.845 [289/16/4256] Linking CXX static library lib/libLLVMInstrumentation.a
27.928 [288/16/4257] Linking CXX static library lib/libLLVMAggressiveInstCombine.a
28.149 [287/16/4258] Linking CXX static library lib/libLLVMInstCombine.a
28.611 [286/16/4259] Linking CXX static library lib/libLLVMScalarOpts.a
28.725 [285/16/4260] Linking CXX static library lib/libLLVMFrontendOpenMP.a
28.994 [284/16/4261] Linking CXX static library lib/libLLVMVectorize.a
29.095 [283/16/4262] Linking CXX static library lib/libLLVMObjCARCOpts.a
29.175 [282/16/4263] Linking CXX static library lib/libLLVMHipStdPar.a
29.251 [281/16/4264] Building CXX object tools/mlir/lib/Conversion/VectorToXeGPU/CMakeFiles/obj.MLIRVectorToXeGPU.dir/VectorToXeGPU.cpp.o
FAILED: tools/mlir/lib/Conversion/VectorToXeGPU/CMakeFiles/obj.MLIRVectorToXeGPU.dir/VectorToXeGPU.cpp.o 
CCACHE_CPP2=yes CCACHE_HASHDIR=yes /usr/bin/ccache /usr/bin/g++-7 -DGTEST_HAS_RTTI=0 -D_DEBUG -D_GLIBCXX_ASSERTIONS -D_GNU_SOURCE -D__STDC_CONSTANT_MACROS -D__STDC_FORMAT_MACROS -D__STDC_LIMIT_MACROS -I/vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.obj/tools/mlir/lib/Conversion/VectorToXeGPU -I/vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.src/mlir/lib/Conversion/VectorToXeGPU -I/vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.obj/include -I/vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.src/llvm/include -I/vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.src/mlir/include -I/vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.obj/tools/mlir/include -fPIC -fno-semantic-interposition -fvisibility-inlines-hidden -Werror=date-time -fno-lifetime-dse -Wall -Wextra -Wno-unused-parameter -Wwrite-strings -Wcast-qual -Wno-missing-field-initializers -pedantic -Wno-long-long -Wimplicit-fallthrough -Wno-uninitialized -Wno-nonnull -Wno-noexcept-type -Wdelete-non-virtual-dtor -Wno-comment -Wno-misleading-indentation -fdiagnostics-color -ffunction-sections -fdata-sections -Wundef -Wno-unused-but-set-parameter -O3 -DNDEBUG  -fno-exceptions -funwind-tables -fno-rtti -UNDEBUG -std=c++1z -MD -MT tools/mlir/lib/Conversion/VectorToXeGPU/CMakeFiles/obj.MLIRVectorToXeGPU.dir/VectorToXeGPU.cpp.o -MF tools/mlir/lib/Conversion/VectorToXeGPU/CMakeFiles/obj.MLIRVectorToXeGPU.dir/VectorToXeGPU.cpp.o.d -o tools/mlir/lib/Conversion/VectorToXeGPU/CMakeFiles/obj.MLIRVectorToXeGPU.dir/VectorToXeGPU.cpp.o -c /vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.src/mlir/lib/Conversion/VectorToXeGPU/VectorToXeGPU.cpp
In file included from /vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.src/mlir/include/mlir/Dialect/XeGPU/IR/XeGPU.h:34:0,
                 from /vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.src/mlir/lib/Conversion/VectorToXeGPU/VectorToXeGPU.cpp:18:
/vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.obj/tools/mlir/include/mlir/Dialect/XeGPU/IR/XeGPU.h.inc: In member function ‘llvm::ArrayRef<long int> mlir::xegpu::CreateNdDescOp::getStaticStrides()’:
/vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.obj/tools/mlir/include/mlir/Dialect/XeGPU/IR/XeGPU.h.inc:1201:26: warning: unused variable ‘offset’ [-Wunused-variable]
     auto [strides, offset] = getStridesAndOffset(memrefType);
                          ^
/vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.src/mlir/lib/Conversion/VectorToXeGPU/VectorToXeGPU.cpp: In function ‘mlir::xegpu::CreateNdDescOp {anonymous}::createNdDescriptor(mlir::PatternRewriter&, mlir::Location, mlir::xegpu::TensorDescType, mlir::TypedValue<mlir::MemRefType>, mlir::Operation::operand_range)’:
/vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.src/mlir/lib/Conversion/VectorToXeGPU/VectorToXeGPU.cpp:89:24: warning: unused variable ‘offset’ [-Wunused-variable]
   auto [strides, offset] = getStridesAndOffset(srcTy);
                        ^
/vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.src/mlir/lib/Conversion/VectorToXeGPU/VectorToXeGPU.cpp: In member function ‘virtual llvm::LogicalResult {anonymous}::TransferReadLowering::matchAndRewrite(mlir::vector::TransferReadOp, mlir::PatternRewriter&) const’:
/vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.src/mlir/lib/Conversion/VectorToXeGPU/VectorToXeGPU.cpp:172:16: error: ‘mlir::xegpu::MemoryScope’ has not been declared
         xegpu::MemoryScope::Global,
                ^~~~~~~~~~~
/vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.src/mlir/lib/Conversion/VectorToXeGPU/VectorToXeGPU.cpp: In member function ‘virtual llvm::LogicalResult {anonymous}::TransferWriteLowering::matchAndRewrite(mlir::vector::TransferWriteOp, mlir::PatternRewriter&) const’:
/vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.src/mlir/lib/Conversion/VectorToXeGPU/VectorToXeGPU.cpp:217:57: error: ‘mlir::xegpu::MemoryScope’ has not been declared
         /*scattered=*/false, /*array_length=*/1, xegpu::MemoryScope::Global,
                                                         ^~~~~~~~~~~
/vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.src/mlir/lib/Conversion/VectorToXeGPU/VectorToXeGPU.cpp: At global scope:
/vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.src/mlir/lib/Conversion/VectorToXeGPU/VectorToXeGPU.cpp:85:1: warning: ‘mlir::xegpu::CreateNdDescOp {anonymous}::createNdDescriptor(mlir::PatternRewriter&, mlir::Location, mlir::xegpu::TensorDescType, mlir::TypedValue<mlir::MemRefType>, mlir::Operation::operand_range)’ defined but not used [-Wunused-function]
 createNdDescriptor(PatternRewriter &rewriter, Location loc,
 ^~~~~~~~~~~~~~~~~~
29.268 [281/15/4265] Linking CXX static library lib/libLLVMLinker.a
30.105 [281/14/4266] Building CXX object tools/mlir/lib/Dialect/XeGPU/Transforms/CMakeFiles/obj.MLIRXeGPUTransforms.dir/XeGPUFoldAliasOps.cpp.o
In file included from /vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.src/mlir/include/mlir/Dialect/XeGPU/IR/XeGPU.h:34:0,
                 from /vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.src/mlir/lib/Dialect/XeGPU/Transforms/XeGPUFoldAliasOps.cpp:13:
/vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.obj/tools/mlir/include/mlir/Dialect/XeGPU/IR/XeGPU.h.inc: In member function ‘llvm::ArrayRef<long int> mlir::xegpu::CreateNdDescOp::getStaticStrides()’:
/vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.obj/tools/mlir/include/mlir/Dialect/XeGPU/IR/XeGPU.h.inc:1201:26: warning: unused variable ‘offset’ [-Wunused-variable]
     auto [strides, offset] = getStridesAndOffset(memrefType);
                          ^
35.691 [281/13/4267] Building CXX object tools/mlir/lib/Dialect/XeGPU/IR/CMakeFiles/obj.MLIRXeGPUDialect.dir/XeGPUDialect.cpp.o
In file included from /vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.src/mlir/include/mlir/Dialect/XeGPU/IR/XeGPU.h:34:0,
                 from /vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.src/mlir/lib/Dialect/XeGPU/IR/XeGPUDialect.cpp:9:
/vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.obj/tools/mlir/include/mlir/Dialect/XeGPU/IR/XeGPU.h.inc: In member function ‘llvm::ArrayRef<long int> mlir::xegpu::CreateNdDescOp::getStaticStrides()’:
/vol/worker/mlir-nvidia/mlir-nvidia-gcc7/llvm.obj/tools/mlir/include/mlir/Dialect/XeGPU/IR/XeGPU.h.inc:1201:26: warning: unused variable ‘offset’ [-Wunused-variable]
     auto [strides, offset] = getStridesAndOffset(memrefType);
                          ^

```

</details>

https://github.com/llvm/llvm-project/pull/109144

More information about the Mlir-commits mailing list