Skip to content

Commit

Permalink
Code cleanup and some upgrade to builders (#2920)
Browse files Browse the repository at this point in the history 
Signed-off-by: Alexandre Eichenberger <[email protected]>
  • Loading branch information
@AlexandreEichenberger
AlexandreEichenberger authored Aug 28, 2024
1 parent 4f0a141 commit f7d8db5
Show file tree
Hide file tree
Showing 159 changed files with 1,090 additions and 1,080 deletions.
2 changes: 1 addition & 1 deletion src/Accelerators/NNPA/Conversion/ONNXToZHigh/ONNXToZHigh.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -207,7 +207,7 @@ Value getLSTMGRUGetYc(
SmallVector<Value, 4> emitONNXSplitOp(Location loc, PatternRewriter &rewriter,
Value input, IntegerAttr axis, ArrayAttr split) {
Type elementType = mlir::cast<ShapedType>(input.getType()).getElementType();
SmallVector<mlir::Type> outputTypes;
SmallVector<Type> outputTypes;
int64_t splitNum = split.size();
ArrayRef<int64_t> inputShape =
mlir::cast<RankedTensorType>(input.getType()).getShape();
Expand Down
6 changes: 3 additions & 3 deletions src/Accelerators/NNPA/Conversion/ONNXToZHigh/ONNXToZHighCommon.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -73,10 +73,10 @@ ValueRange splitAlongAxis(
return splits;
}

bool isF32ScalarConstantTensor(mlir::Value v) {
bool isF32ScalarConstantTensor(Value v) {
if (!isScalarConstantTensor(v))
return false;
auto t = dyn_cast<ShapedType>(v.getType());
auto t = mlir::dyn_cast<ShapedType>(v.getType());
return t.getElementType().isF32();
}

Expand All @@ -93,7 +93,7 @@ Value getDynShape(Location loc, PatternRewriter &rewriter, Value x) {
llvm_unreachable("The input must have shape and rank");

OnnxBuilder create(rewriter, loc);
auto t = dyn_cast<ShapedType>(x.getType());
auto t = mlir::dyn_cast<ShapedType>(x.getType());
int64_t r = t.getRank();
SmallVector<Value> dims;
for (int64_t i = 0; i < r; ++i) {
Expand Down
32 changes: 16 additions & 16 deletions src/Accelerators/NNPA/Conversion/ONNXToZHigh/PerfModel.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -416,44 +416,44 @@ double estimateTimeForUnstickOp(Value oper) {
bool estimateTimeForOpWithModel(Operation *op, const DimAnalysis *dimAnalysis,
double &cpuEstimatedTime, double &nnpaEstimatedTime) {
bool opHasModel = true;
if (auto addOp = dyn_cast<ONNXAddOp>(op))
if (auto addOp = mlir::dyn_cast<ONNXAddOp>(op))
estimateTimeForOp(addOp, dimAnalysis, cpuEstimatedTime, nnpaEstimatedTime);
else if (auto divOp = dyn_cast<ONNXDivOp>(op))
else if (auto divOp = mlir::dyn_cast<ONNXDivOp>(op))
estimateTimeForOp(divOp, dimAnalysis, cpuEstimatedTime, nnpaEstimatedTime);
else if (auto maxOp = dyn_cast<ONNXMaxOp>(op))
else if (auto maxOp = mlir::dyn_cast<ONNXMaxOp>(op))
estimateTimeForOp(maxOp, dimAnalysis, cpuEstimatedTime, nnpaEstimatedTime);
else if (auto minOp = dyn_cast<ONNXMinOp>(op))
else if (auto minOp = mlir::dyn_cast<ONNXMinOp>(op))
estimateTimeForOp(minOp, dimAnalysis, cpuEstimatedTime, nnpaEstimatedTime);
else if (auto mulOp = dyn_cast<ONNXMulOp>(op))
else if (auto mulOp = mlir::dyn_cast<ONNXMulOp>(op))
estimateTimeForOp(mulOp, dimAnalysis, cpuEstimatedTime, nnpaEstimatedTime);
else if (auto powOp = dyn_cast<ONNXPowOp>(op))
else if (auto powOp = mlir::dyn_cast<ONNXPowOp>(op))
estimateTimeForOp(powOp, dimAnalysis, cpuEstimatedTime, nnpaEstimatedTime);
else if (auto subOp = dyn_cast<ONNXSubOp>(op))
else if (auto subOp = mlir::dyn_cast<ONNXSubOp>(op))
estimateTimeForOp(subOp, dimAnalysis, cpuEstimatedTime, nnpaEstimatedTime);
// Unary elementwise NNPA candidate ops.
else if (auto expOp = dyn_cast<ONNXExpOp>(op))
else if (auto expOp = mlir::dyn_cast<ONNXExpOp>(op))
estimateTimeForOp(expOp, dimAnalysis, cpuEstimatedTime, nnpaEstimatedTime);
else if (auto logOp = dyn_cast<ONNXLogOp>(op))
else if (auto logOp = mlir::dyn_cast<ONNXLogOp>(op))
estimateTimeForOp(logOp, dimAnalysis, cpuEstimatedTime, nnpaEstimatedTime);
else if (auto reluOp = dyn_cast<ONNXReluOp>(op))
else if (auto reluOp = mlir::dyn_cast<ONNXReluOp>(op))
estimateTimeForOp(reluOp, dimAnalysis, cpuEstimatedTime, nnpaEstimatedTime);
else if (auto sigmoidOp = dyn_cast<ONNXSigmoidOp>(op))
else if (auto sigmoidOp = mlir::dyn_cast<ONNXSigmoidOp>(op))
estimateTimeForOp(
sigmoidOp, dimAnalysis, cpuEstimatedTime, nnpaEstimatedTime);
else if (auto softmaxOp = dyn_cast<ONNXSoftmaxOp>(op))
else if (auto softmaxOp = mlir::dyn_cast<ONNXSoftmaxOp>(op))
estimateTimeForOp(
softmaxOp, dimAnalysis, cpuEstimatedTime, nnpaEstimatedTime);
else if (auto tanhOp = dyn_cast<ONNXTanhOp>(op))
else if (auto tanhOp = mlir::dyn_cast<ONNXTanhOp>(op))
estimateTimeForOp(tanhOp, dimAnalysis, cpuEstimatedTime, nnpaEstimatedTime);
// Reduce
else if (auto reduceMeanOp = dyn_cast<ONNXReduceMeanV13Op>(op))
else if (auto reduceMeanOp = mlir::dyn_cast<ONNXReduceMeanV13Op>(op))
estimateTimeForOp(
reduceMeanOp, dimAnalysis, cpuEstimatedTime, nnpaEstimatedTime);
// Matmul.
else if (auto matMulOp = dyn_cast<ONNXMatMulOp>(op))
else if (auto matMulOp = mlir::dyn_cast<ONNXMatMulOp>(op))
estimateTimeForOp(
matMulOp, dimAnalysis, cpuEstimatedTime, nnpaEstimatedTime);
else if (auto gemmOp = dyn_cast<ONNXGemmOp>(op))
else if (auto gemmOp = mlir::dyn_cast<ONNXGemmOp>(op))
estimateTimeForOp(gemmOp, dimAnalysis, cpuEstimatedTime, nnpaEstimatedTime);
else
opHasModel = false;
Expand Down
4 changes: 2 additions & 2 deletions src/Accelerators/NNPA/Conversion/ONNXToZHigh/RewriteONNXForZHigh.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -219,7 +219,7 @@ bool canInferencePadsForNNPAConv(ONNXConvOp op) {
// Create an ArrayAttr of IntegerAttr(s) of zero values.
// This function is used for padding attribute in Conv.
ArrayAttr getPadsForNNPAConv(PatternRewriter &rewriter, Value ret) {
ONNXConvOp op = dyn_cast<ONNXConvOp>(ret.getDefiningOp());
ONNXConvOp op = mlir::dyn_cast<ONNXConvOp>(ret.getDefiningOp());
ONNXConvOpShapeHelper shapeHelper(op.getOperation(), {});
shapeHelper.computeShapeAndAssertOnFailure();
SmallVector<int64_t, 4> vals;
Expand Down Expand Up @@ -451,7 +451,7 @@ class AddSubWithRHSZeroExpandPattern : public OpRewritePattern<OP_TYPE> {
if (isa<BlockArgument>(B))
return false;
bool BIsZero = false;
if (auto expandOp = dyn_cast<ONNXExpandOp>(B.getDefiningOp())) {
if (auto expandOp = mlir::dyn_cast<ONNXExpandOp>(B.getDefiningOp())) {
Value input = expandOp.getInput();
if (isDenseONNXConstant(input)) {
// Expand's input is 0?
Expand Down
4 changes: 2 additions & 2 deletions src/Accelerators/NNPA/Conversion/ZHighToZLow/ZHighToZLow.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -501,7 +501,7 @@ struct ZHighToZLowStickOpLowering : public ConversionPattern {
LogicalResult matchAndRewrite(Operation *op, ArrayRef<Value> operands,
ConversionPatternRewriter &rewriter) const final {
Location loc = op->getLoc();
ZHighStickOp stickOp = cast<ZHighStickOp>(op);
ZHighStickOp stickOp = mlir::cast<ZHighStickOp>(op);

ZHighStickOpAdaptor operandAdaptor(operands);
Value input = operandAdaptor.getIn();
Expand Down Expand Up @@ -1557,7 +1557,7 @@ struct ZHighToZLowStickifiedConstantOfShapeOpLowering
Location loc = op->getLoc();
MDBuilder create(rewriter, loc);

auto stickOp = cast<ZHighStickifiedConstantOfShapeOp>(op);
auto stickOp = mlir::cast<ZHighStickifiedConstantOfShapeOp>(op);
FloatAttr value = stickOp.getValueAttr();
Type i16Ty = rewriter.getI16Type();
Type i64Ty = rewriter.getI64Type();
Expand Down
28 changes: 14 additions & 14 deletions src/Accelerators/NNPA/Conversion/ZLowToLLVM/ZLowToLLVM.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -98,7 +98,7 @@ class ZLowStickLowering : public mlir::ConvertToLLVMPattern {
ConversionPatternRewriter &rewriter) const override {
ModuleOp module = op->getParentOfType<ModuleOp>();
Location loc = op->getLoc();
ZLowStickOp stickOp = cast<ZLowStickOp>(op);
ZLowStickOp stickOp = mlir::cast<ZLowStickOp>(op);

ZLowStickOpAdaptor operandAdaptor(operands);
// Do not get element type from adaptor since the type can be opaque.
Expand Down Expand Up @@ -154,7 +154,7 @@ class ZLowStickForLSTMLowering : public ConvertToLLVMPattern {
ConversionPatternRewriter &rewriter) const override {
ModuleOp module = op->getParentOfType<ModuleOp>();
Location loc = op->getLoc();
ZLowStickForLSTMOp stickForLSTMOp = cast<ZLowStickForLSTMOp>(op);
ZLowStickForLSTMOp stickForLSTMOp = mlir::cast<ZLowStickForLSTMOp>(op);

ZLowStickForLSTMOpAdaptor operandAdaptor(operands);
Type llvmElementTy = typeConverter->convertType(
Expand Down Expand Up @@ -240,7 +240,7 @@ class ZLowStickForGRULowering : public ConvertToLLVMPattern {
ConversionPatternRewriter &rewriter) const override {
ModuleOp module = op->getParentOfType<ModuleOp>();
Location loc = op->getLoc();
ZLowStickForGRUOp stickForGRUOp = cast<ZLowStickForGRUOp>(op);
ZLowStickForGRUOp stickForGRUOp = mlir::cast<ZLowStickForGRUOp>(op);

ZLowStickForGRUOpAdaptor operandAdaptor(operands);
Type llvmElementTy = typeConverter->convertType(
Expand Down Expand Up @@ -324,7 +324,7 @@ class ZLowLSTMLowering : public ConvertToLLVMPattern {
ConversionPatternRewriter &rewriter) const override {
ModuleOp module = op->getParentOfType<ModuleOp>();
Location loc = op->getLoc();
ZLowLSTMOp lstmOp = cast<ZLowLSTMOp>(op);
ZLowLSTMOp lstmOp = mlir::cast<ZLowLSTMOp>(op);
MultiDialectBuilder<LLVMBuilder> create(rewriter, loc);

ZLowLSTMOpAdaptor operandAdaptor(operands);
Expand Down Expand Up @@ -520,7 +520,7 @@ class ZLowGRULowering : public ConvertToLLVMPattern {
ConversionPatternRewriter &rewriter) const override {
ModuleOp module = op->getParentOfType<ModuleOp>();
Location loc = op->getLoc();
ZLowGRUOp gruOp = cast<ZLowGRUOp>(op);
ZLowGRUOp gruOp = mlir::cast<ZLowGRUOp>(op);
MultiDialectBuilder<LLVMBuilder> create(rewriter, loc);

ZLowGRUOpAdaptor operandAdaptor(operands);
Expand Down Expand Up @@ -675,7 +675,7 @@ class ZLowUnstickLowering : public ConvertToLLVMPattern {
ConversionPatternRewriter &rewriter) const override {
ModuleOp module = op->getParentOfType<ModuleOp>();
Location loc = op->getLoc();
ZLowUnstickOp unstickOp = cast<ZLowUnstickOp>(op);
ZLowUnstickOp unstickOp = mlir::cast<ZLowUnstickOp>(op);

ZLowUnstickOpAdaptor operandAdaptor(operands);
Type llvmElementTy = typeConverter->convertType(
Expand Down Expand Up @@ -732,7 +732,7 @@ class ZLowUnaryElementwiseOpLowering : public ConvertToLLVMPattern {
ModuleOp module = op->getParentOfType<ModuleOp>();
Location loc = op->getLoc();
MLIRContext *context = rewriter.getContext();
UnaryElementwiseOp unaryOp = cast<UnaryElementwiseOp>(op);
UnaryElementwiseOp unaryOp = mlir::cast<UnaryElementwiseOp>(op);
typename UnaryElementwiseOp::Adaptor operandAdaptor(operands);
MultiDialectBuilder<LLVMBuilder> create(rewriter, loc);

Expand Down Expand Up @@ -810,7 +810,7 @@ class ZLowBinaryElementwiseOpLowering : public ConvertToLLVMPattern {
ConversionPatternRewriter &rewriter) const override {
ModuleOp module = op->getParentOfType<ModuleOp>();
Location loc = op->getLoc();
BinaryElementwiseOp binaryOp = cast<BinaryElementwiseOp>(op);
BinaryElementwiseOp binaryOp = mlir::cast<BinaryElementwiseOp>(op);
typename BinaryElementwiseOp::Adaptor operandAdaptor(operands);

Value input1 = operandAdaptor.getX();
Expand Down Expand Up @@ -888,7 +888,7 @@ class ZLowSoftmaxOpLowering : public ConvertToLLVMPattern {
ConversionPatternRewriter &rewriter) const override {
ModuleOp module = op->getParentOfType<ModuleOp>();
Location loc = op->getLoc();
ZLowSoftmaxOp softmaxOp = cast<ZLowSoftmaxOp>(op);
ZLowSoftmaxOp softmaxOp = mlir::cast<ZLowSoftmaxOp>(op);
MultiDialectBuilder<LLVMBuilder> create(rewriter, loc);

ZLowSoftmaxOpAdaptor operandAdaptor(operands);
Expand Down Expand Up @@ -971,7 +971,7 @@ class ZLowMatMulLowering : public ConvertToLLVMPattern {
ConversionPatternRewriter &rewriter) const override {
ModuleOp module = op->getParentOfType<ModuleOp>();
Location loc = op->getLoc();
ZLowMatMulOp matmulOp = cast<ZLowMatMulOp>(op);
ZLowMatMulOp matmulOp = mlir::cast<ZLowMatMulOp>(op);
MultiDialectBuilder<LLVMBuilder> create(rewriter, loc);

ZLowMatMulOpAdaptor operandAdaptor(operands);
Expand Down Expand Up @@ -1109,7 +1109,7 @@ class ZLowConv2DLowering : public ConvertToLLVMPattern {
ModuleOp module = op->getParentOfType<ModuleOp>();
Location loc = op->getLoc();
MLIRContext *context = rewriter.getContext();
ZLowConv2DOp convOp = cast<ZLowConv2DOp>(op);
ZLowConv2DOp convOp = mlir::cast<ZLowConv2DOp>(op);
ZLowConv2DOpAdaptor operandAdaptor(operands);
MultiDialectBuilder<LLVMBuilder> create(rewriter, loc);

Expand Down Expand Up @@ -1256,7 +1256,7 @@ class ZLowPool2DLowering : public ConvertToLLVMPattern {
ConversionPatternRewriter &rewriter) const override {
ModuleOp module = op->getParentOfType<ModuleOp>();
Location loc = op->getLoc();
POOLOP poolOp = cast<POOLOP>(op);
POOLOP poolOp = mlir::cast<POOLOP>(op);
typename POOLOP::Adaptor operandAdaptor(operands);
MultiDialectBuilder<LLVMBuilder> create(rewriter, loc);

Expand Down Expand Up @@ -1360,7 +1360,7 @@ class ZLowMeanReduce2DLowering : public ConvertToLLVMPattern {
ConversionPatternRewriter &rewriter) const override {
ModuleOp module = op->getParentOfType<ModuleOp>();
Location loc = op->getLoc();
ZLowMeanReduce2DOp meanOp = cast<ZLowMeanReduce2DOp>(op);
ZLowMeanReduce2DOp meanOp = mlir::cast<ZLowMeanReduce2DOp>(op);
MultiDialectBuilder<LLVMBuilder> create(rewriter, loc);

ZLowMeanReduce2DOpAdaptor operandAdaptor(operands);
Expand Down Expand Up @@ -1429,7 +1429,7 @@ class ZLowBatchNormLowering : public ConvertToLLVMPattern {
ConversionPatternRewriter &rewriter) const override {
ModuleOp module = op->getParentOfType<ModuleOp>();
Location loc = op->getLoc();
ZLowBatchNormOp batchnormOp = cast<ZLowBatchNormOp>(op);
ZLowBatchNormOp batchnormOp = mlir::cast<ZLowBatchNormOp>(op);

ZLowBatchNormOpAdaptor operandAdaptor(operands);
Type llvmElementTy = typeConverter->convertType(
Expand Down
2 changes: 1 addition & 1 deletion src/Accelerators/NNPA/Conversion/ZLowToLLVM/ZLowToLLVMCommon.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -364,7 +364,7 @@ std::vector<Value> getDimsFromShapeMemRefBySize(PatternRewriter &rewriter,
bitcastOp.getArg().getDefiningOp());
if (addressOfOp) {
LLVM::GlobalOp globalOp =
dyn_cast_or_null<LLVM::GlobalOp>(SymbolTable::lookupSymbolIn(
mlir::dyn_cast_or_null<LLVM::GlobalOp>(SymbolTable::lookupSymbolIn(
module, addressOfOp.getGlobalNameAttr()));
if (globalOp) {
DenseElementsAttr valueAttr =
Expand Down
4 changes: 2 additions & 2 deletions src/Accelerators/NNPA/Dialect/ZHigh/ZHighOps.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -71,10 +71,10 @@ LogicalResult OpTrait::impl::verifySameOperandsAndResultLayout(Operation *op) {
namespace onnx_mlir {
namespace zhigh {

std::vector<mlir::Type> getZHighAuxSplitResultType(
std::vector<Type> getZHighAuxSplitResultType(
Value input, int64_t axis, ArrayAttr split) {
Type elementType = mlir::cast<ShapedType>(input.getType()).getElementType();
std::vector<mlir::Type> outputTypes;
std::vector<Type> outputTypes;
if (split.size() == 0) {
llvm_unreachable("Unsupported split (size==0)");
} else {
Expand Down
2 changes: 1 addition & 1 deletion src/Accelerators/NNPA/Dialect/ZHigh/ZHighOps/BatchNorm/BatchNorm.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -24,7 +24,7 @@ namespace zhigh {
//===----------------------------------------------------------------------===//

LogicalResult ZHighBatchNormOp::inferShapes(
std::function<void(mlir::Region &)> doShapeInference) {
std::function<void(Region &)> doShapeInference) {
return inferShapeForUnaryOps(this->getOperation());
}

Expand Down
2 changes: 1 addition & 1 deletion src/Accelerators/NNPA/Dialect/ZHigh/ZHighOps/Conv2D/Conv2D.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -139,7 +139,7 @@ LogicalResult ZHighConv2DOp::verify() {
//===----------------------------------------------------------------------===//

LogicalResult ZHighConv2DOp::inferShapes(
std::function<void(mlir::Region &)> doShapeInference) {
std::function<void(Region &)> doShapeInference) {
if (!hasRankedType(getInput()) || !hasRankedType(getInputKernel()))
return success();

Expand Down
4 changes: 2 additions & 2 deletions src/Accelerators/NNPA/Dialect/ZHigh/ZHighOps/DLF16ToF32/DLF16ToF32.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -33,7 +33,7 @@ void ZHighDLF16ToF32Op::build(
Type elementType = builder.getF32Type();
Type resType = UnrankedTensorType::get(elementType);

if (auto inType = dyn_cast<RankedTensorType>(input.getType()))
if (auto inType = mlir::dyn_cast<RankedTensorType>(input.getType()))
resType = RankedTensorType::get(inType.getShape(), elementType);

build(builder, state, resType, input);
Expand All @@ -44,7 +44,7 @@ void ZHighDLF16ToF32Op::build(
//===----------------------------------------------------------------------===//

LogicalResult ZHighDLF16ToF32Op::inferShapes(
std::function<void(mlir::Region &)> doShapeInference) {
std::function<void(Region &)> doShapeInference) {
return inferShapeForUnaryOps(this->getOperation());
}

Expand Down
Loading

0 comments on commit f7d8db5

Please sign in to comment.