llvm
/
torch-mlir
mirror of https://github.com/llvm/torch-mlir
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

[NFC reformat] Run pre-commit on all files and format misc. 

This is part 1 of ~3, formatting all miscellaneous text files and CPP files matched by a first run of pre-commit. These tend to be low change-traffic and are likely not disruptive.

Subsequent patches will format Python files and remaining CPP files.
pull/3244/head
Stella Laurenzo 2024-04-27 14:08:09 -07:00
parent 6679728c56
commit 5d4b803914
40 changed files with 99 additions and 113 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
docs/importers/onnx_importer.md
View File

@ -140,4 +140,3 @@ torch-mlir's representation:
* `ConstantOfShape`: Mapped to `torch.vtensor.literal` with
a corresponding `value` attribute.

1
docs/roadmap.md
View File

@ -277,4 +277,3 @@ directly provided a way to plug into this.
Additionally, we can leverage the [`pytorch-jit-paritybench`](https://github.com/jansel/pytorch-jit-paritybench)
to verify our end-to-end correctness on real models.

65
lib/Conversion/TorchOnnxToTorch/DefaultDomainGtoP.cpp
View File

@ -628,42 +628,39 @@ void mlir::torch::onnx_c::populateDefaultDomainGtoP(
binder.op, resultType, operand);
return success();
});
patterns.onOp("Not", 1,
[](OpBinder binder, ConversionPatternRewriter &rewriter) {
Torch::ValueTensorType resultType;
Value operand;
if (binder.tensorOperand(operand) ||
binder.tensorResultType(resultType)) {
return failure();
}
patterns.onOp(
"Not", 1, [](OpBinder binder, ConversionPatternRewriter &rewriter) {
Torch::ValueTensorType resultType;
Value operand;
if (binder.tensorOperand(operand) ||
binder.tensorResultType(resultType)) {
return failure();
}
auto loc = binder.getLoc();
auto operandTy =
cast<Torch::ValueTensorType>(operand.getType());
auto eTy = operandTy.getDtype();
auto loc = binder.getLoc();
auto operandTy = cast<Torch::ValueTensorType>(operand.getType());
auto eTy = operandTy.getDtype();
if (!eTy.isInteger(1)) {
auto i1ty = rewriter.getI1Type();
auto ty = rewriter.getType<Torch::ValueTensorType>(
operandTy.getSizes(), i1ty);
auto torchqTy = Torch::getScalarTypeForType(i1ty);
Value tyConst = rewriter.create<Torch::ConstantIntOp>(
binder.getLoc(), rewriter.getType<Torch::IntType>(),
rewriter.getIntegerAttr(
rewriter.getIntegerType(64),
static_cast<int64_t>(torchqTy)));
Value none = rewriter.create<Torch::ConstantNoneOp>(loc);
Value cstFalse =
rewriter.create<Torch::ConstantBoolOp>(loc, false);
operand = rewriter.create<Torch::AtenToDtypeOp>(
loc, ty, operand, tyConst,
/*non_blocking=*/cstFalse, /*copy=*/cstFalse,
/*memory_format=*/none);
}
rewriter.replaceOpWithNewOp<Torch::AtenBitwiseNotOp>(
binder.op, resultType, operand);
return success();
});
if (!eTy.isInteger(1)) {
auto i1ty = rewriter.getI1Type();
auto ty = rewriter.getType<Torch::ValueTensorType>(
operandTy.getSizes(), i1ty);
auto torchqTy = Torch::getScalarTypeForType(i1ty);
Value tyConst = rewriter.create<Torch::ConstantIntOp>(
binder.getLoc(), rewriter.getType<Torch::IntType>(),
rewriter.getIntegerAttr(rewriter.getIntegerType(64),
static_cast<int64_t>(torchqTy)));
Value none = rewriter.create<Torch::ConstantNoneOp>(loc);
Value cstFalse = rewriter.create<Torch::ConstantBoolOp>(loc, false);
operand = rewriter.create<Torch::AtenToDtypeOp>(
loc, ty, operand, tyConst,
/*non_blocking=*/cstFalse, /*copy=*/cstFalse,
/*memory_format=*/none);
}
rewriter.replaceOpWithNewOp<Torch::AtenBitwiseNotOp>(
binder.op, resultType, operand);
return success();
});
patterns.onOp("Or", 1,
[](OpBinder binder, ConversionPatternRewriter &rewriter) {
Torch::ValueTensorType resultType;

5
lib/Conversion/TorchToStablehlo/StablehloLegalizeUtils.cpp
View File

@ -189,9 +189,8 @@ Value promoteAndBroadcast(ConversionPatternRewriter &rewriter, Value input,
do_bcast = true;
} else {
op->emitError("The size of tensor a (")
<< inDim << ")"
<< "must match the size of tensor b (" << outDim << ")"
<< "at non-singleton dimension " << inPos;
<< inDim << ")" << "must match the size of tensor b (" << outDim
<< ")" << "at non-singleton dimension " << inPos;
}
}
std::reverse(bcastDims.begin(), bcastDims.end());

3
lib/Dialect/Torch/Transforms/LowerToBackendContract.cpp
View File

@ -305,8 +305,7 @@ public:
return signalPassFailure();
} while (!satisfiesBackendContract(module, target));
LLVM_DEBUG({
llvm::dbgs() << "LowerToBackendContractPass: "
<< "succeeded after " << i
llvm::dbgs() << "LowerToBackendContractPass: " << "succeeded after " << i
<< " iterations of the simplification pipeline\n";
});
}

1
projects/pt1/python/torch_mlir/_torch_mlir_custom_op_example/CMakeLists.txt
View File

@ -28,4 +28,3 @@ set_target_properties(torch_mlir_custom_op_example PROPERTIES
)
torch_mlir_python_target_compile_options(torch_mlir_custom_op_example)
mlir_check_all_link_libraries(torch_mlir_custom_op_example)

14
test/CAPI/torch.c
View File

@ -36,7 +36,7 @@ static void testTensor(MlirContext ctx, intptr_t numSizes, int64_t *sizes,
fprintf(stderr, #TTT "Type %s rank: %zu\n", testName, \
torchMlirTorch##TTT##TypeGetRank(TTT##Type)); \
int64_t *TTT##Sizes = malloc(sizeof(int64_t) * numSizes); \
torchMlirTorch##TTT##TypeGetSizes(TTT##Type, TTT##Sizes); \
torchMlirTorch##TTT##TypeGetSizes(TTT##Type, TTT##Sizes); \
for (int i = 0; i < numSizes; ++i) { \
fprintf(stderr, #TTT "Type %s pos %d size: %ld\n", testName, i, \
TTT##Sizes[i]); \
@ -157,22 +157,26 @@ static void testTypeMetaDataAccessors(MlirContext ctx) {
MlirType dictType1 = torchMlirTorchDictTypeGet(strType, floatType);
fprintf(stderr, "dict keyType: ");
mlirTypePrint(torchMlirTorchDictTypeGetKeyType(dictType1), printToStderr, NULL);
mlirTypePrint(torchMlirTorchDictTypeGetKeyType(dictType1), printToStderr,
NULL);
fprintf(stderr, "\n");
// CHECK: dict keyType: !torch.str
fprintf(stderr, "dict valueType: ");
mlirTypePrint(torchMlirTorchDictTypeGetValueType(dictType1), printToStderr, NULL);
mlirTypePrint(torchMlirTorchDictTypeGetValueType(dictType1), printToStderr,
NULL);
fprintf(stderr, "\n");
// CHECK: dict valueType: !torch.float
MlirType dictType2 = torchMlirTorchDictTypeGet(floatType, strType);
fprintf(stderr, "dict keyType: ");
mlirTypePrint(torchMlirTorchDictTypeGetKeyType(dictType2), printToStderr, NULL);
mlirTypePrint(torchMlirTorchDictTypeGetKeyType(dictType2), printToStderr,
NULL);
fprintf(stderr, "\n");
// CHECK: dict keyType: !torch.float
fprintf(stderr, "dict valueType: ");
mlirTypePrint(torchMlirTorchDictTypeGetValueType(dictType2), printToStderr, NULL);
mlirTypePrint(torchMlirTorchDictTypeGetValueType(dictType2), printToStderr,
NULL);
fprintf(stderr, "\n");
// CHECK: dict valueType: !torch.str
}

2
test/Conversion/TorchToLinalg/flatten.mlir
View File

@ -82,5 +82,3 @@ func.func @torch.aten.flatten.using_ints$rank0(%arg0: !torch.vtensor<[],f32>) ->
%0 = torch.aten.flatten.using_ints %arg0, %int0, %int0 : !torch.vtensor<[],f32>, !torch.int, !torch.int -> !torch.vtensor<[1],f32>
return %0 : !torch.vtensor<[1],f32>
}

1
test/Conversion/TorchToLinalg/gridsampler.mlir
View File

@ -86,4 +86,3 @@ func.func @grid_sampler3(%arg0: !torch.vtensor<[?,?,?,?],f32>, %arg1: !torch.vte
%4 = torch.aten.grid_sampler %arg0, %arg1, %int0, %int1, %false : !torch.vtensor<[?,?,?,?],f32>, !torch.vtensor<[?,?,?,?],f32>, !torch.int, !torch.int, !torch.bool -> !torch.vtensor<[?,?,?,?],f32>
return %4 : !torch.vtensor<[?,?,?,?],f32>
}

1
test/Conversion/TorchToLinalg/view.mlir
View File

@ -254,4 +254,3 @@ func.func @torch.aten.view$dynamicInferredSame(%arg0: !torch.vtensor<[10,?,2,3],
%1 = torch.aten.view %arg0, %0 : !torch.vtensor<[10,?,2,3],f32>, !torch.list<int> -> !torch.vtensor<[2,5,?,6],f32>
return %1 : !torch.vtensor<[2,5,?,6],f32>
}

1
test/Conversion/TorchToStablehlo/gather.mlir
View File

@ -63,4 +63,3 @@ func.func @torch.aten.embedding$rank_two_indices(%weight: !torch.vtensor<[?,?],f
%ret = torch.aten.embedding %weight, %indices, %int-1, %false, %false : !torch.vtensor<[?,?],f32>, !torch.vtensor<[?,1], si64>, !torch.int, !torch.bool, !torch.bool -> !torch.vtensor<[?,1,?],f32>
return %ret: !torch.vtensor<[?,1,?],f32>
}

1
test/Conversion/TorchToStablehlo/view_like.mlir
View File

@ -565,4 +565,3 @@ func.func @torch.aten.unsqueeze$from_end(%arg0: !torch.vtensor<[?,?,?,?],f32>) -
%0 = torch.aten.unsqueeze %arg0, %int-2 : !torch.vtensor<[?,?,?,?],f32>, !torch.int -> !torch.vtensor<[?,?,?,1,?],f32>
return %0 : !torch.vtensor<[?,?,?,1,?],f32>
}

2
test/Conversion/TorchToTosa/cast_fp32_to_fp16.mlir
View File

@ -8,5 +8,3 @@ func.func @forward(%arg0: !torch.vtensor<[1,32,220,220],f32>) -> !torch.vtensor<
%out = torch.aten.to.dtype %arg0, %int5, %false, %false, %none : !torch.vtensor<[1,32,220,220],f32>, !torch.int, !torch.bool, !torch.bool, !torch.none -> !torch.vtensor<[1,32,220,220],f16>
return %out : !torch.vtensor<[1,32,220,220],f16>
}

1
test/Conversion/TorchToTosa/conv2d_transpose.mlir
View File

@ -15,4 +15,3 @@ func.func @forward(%input: !torch.vtensor<[1,64,1,100],f32>) -> !torch.vtensor<[
%output = torch.aten.convolution %input, %weight, %bias, %stride, %int1x1, %int1x1, %true, %int1x1, %int1 : !torch.vtensor<[1,64,1,100],f32>, !torch.vtensor<[64,64,3,3],f32>, !torch.vtensor<[64],f32>, !torch.list<int>, !torch.list<int>, !torch.list<int>, !torch.bool, !torch.list<int>, !torch.int -> !torch.vtensor<[1,64,2,200],f32>
return %output : !torch.vtensor<[1,64,2,200],f32>
}

1
test/Dialect/Torch/ops.mlir
View File

@ -186,4 +186,3 @@ func.func @torch.permute$negative_index_valid (%arg0: !torch.vtensor<[1,2,3],f32
%3 = torch.aten.permute %arg0, %perm : !torch.vtensor<[1,2,3],f32>, !torch.list<int> -> !torch.vtensor<[1,2,3],f32>
return %3 : !torch.vtensor<[1,2,3],f32>
}