Skip to content

Enhanced splat folding for x+/-0 and x*1 in TOSA #18

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Mar 30, 2023
Merged

Enhanced splat folding for x+/-0 and x*1 in TOSA #18

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 0 additions & 2 deletions mlir/include/mlir/Dialect/Tosa/IR/TosaOps.td
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -419,7 +419,6 @@ def Tosa_AddOp : Tosa_Op<"add", [
Tosa_Tensor:$output
);

let hasCanonicalizer = 1;
let hasFolder = 1;
}

Expand Down Expand Up @@ -738,7 +737,6 @@ def Tosa_MulOp : Tosa_Op<"mul", [
Tosa_Tensor:$output
);

let hasCanonicalizer = 1;
let hasFolder = 1;
}

Expand Down
218 changes: 40 additions & 178 deletions mlir/lib/Dialect/Tosa/IR/TosaCanonicalizations.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -164,92 +164,6 @@ void TransposeOp::getCanonicalizationPatterns(RewritePatternSet &results,
results.add<NoOpOptimization>(context);
}

struct AddZeroOptimization : public OpRewritePattern<tosa::AddOp> {
using OpRewritePattern::OpRewritePattern;

LogicalResult matchAndRewrite(tosa::AddOp op,
PatternRewriter &rewriter) const override {
auto input1 = op.getInput1();
auto input2 = op.getInput2();

DenseElementsAttr input1Attr;
if (matchPattern(input1, m_Constant(&input1Attr)) && input1Attr.isSplat() &&
input2.getType() == op.getType()) {
if (input1Attr.getType().getElementType().isa<IntegerType>() &&
input1Attr.getSplatValue<APInt>().isZero()) {
rewriter.replaceOp(op, op.getInput2());
return success();
}
}

DenseElementsAttr input2Attr;
if (matchPattern(input2, m_Constant(&input2Attr)) && input2Attr.isSplat() &&
input1.getType() == op.getType()) {
if (input2Attr.getType().getElementType().isa<IntegerType>() &&
input2Attr.getSplatValue<APInt>().isZero()) {
rewriter.replaceOp(op, op.getInput1());
return success();
}
}

return failure();
}
};

void AddOp::getCanonicalizationPatterns(RewritePatternSet &results,
MLIRContext *context) {
results.add<AddZeroOptimization>(context);
}

struct MulOneOptimization : public OpRewritePattern<tosa::MulOp> {
using OpRewritePattern::OpRewritePattern;

LogicalResult matchAndRewrite(tosa::MulOp op,
PatternRewriter &rewriter) const override {
auto input1 = op.getInput1();
auto input2 = op.getInput2();

DenseElementsAttr input1Attr;
if (matchPattern(input1, m_Constant(&input1Attr)) && input1Attr.isSplat() &&
input2.getType() == op.getType()) {
if (input1Attr.getType().getElementType().isa<FloatType>() &&
input1Attr.getSplatValue<APFloat>().isExactlyValue(1)) {
rewriter.replaceOp(op, op.getInput2());
return success();
}

if (input1Attr.getType().getElementType().isa<IntegerType>() &&
matchPattern(input1, m_One())) {
rewriter.replaceOp(op, op.getInput2());
return success();
}
}

DenseElementsAttr input2Attr;
if (matchPattern(input2, m_Constant(&input2Attr)) && input2Attr.isSplat() &&
input1.getType() == op.getType()) {
if (input2Attr.getType().getElementType().isa<FloatType>() &&
input2Attr.getSplatValue<APFloat>().isExactlyValue(1)) {
rewriter.replaceOp(op, op.getInput1());
return success();
}

if (input2Attr.getType().getElementType().isa<IntegerType>() &&
matchPattern(input2, m_One())) {
rewriter.replaceOp(op, op.getInput1());
return success();
}
}

return failure();
}
};

void MulOp::getCanonicalizationPatterns(RewritePatternSet &results,
MLIRContext *context) {
results.add<MulOneOptimization>(context);
}

struct MaterializePadValue : public OpRewritePattern<tosa::PadOp> {
using OpRewritePattern::OpRewritePattern;

Expand Down Expand Up @@ -468,64 +382,47 @@ DenseElementsAttr binaryFolder(DenseElementsAttr lhs, DenseElementsAttr rhs,
return {};
}

static bool isSplatZero(Type elemType, DenseElementsAttr val) {
if (elemType.isa<FloatType>())
return val && val.isSplat() && val.getSplatValue<APFloat>().isZero();
if (elemType.isa<IntegerType>())
return val && val.isSplat() && val.getSplatValue<APInt>().isZero();
return false;
}

static bool isSplatOne(Type elemType, DenseElementsAttr val, int64_t shift) {
if (elemType.isa<FloatType>())
return val && val.isSplat() &&
val.getSplatValue<APFloat>().isExactlyValue(1.0);
if (elemType.isa<IntegerType>()) {
const int64_t shifted = 1LL << shift;
return val && val.isSplat() &&
val.getSplatValue<APInt>().getSExtValue() == shifted;
}
return false;
}

OpFoldResult AddOp::fold(ArrayRef<Attribute> operands) {
auto lhsTy = getInput1().getType().dyn_cast<RankedTensorType>();
auto rhsTy = getInput2().getType().dyn_cast<RankedTensorType>();
auto resultTy = getType().dyn_cast<RankedTensorType>();
if (!lhsTy || !rhsTy || !resultTy)
return {};

auto resultETy = resultTy.getElementType();
auto lhsAttr = operands[0].dyn_cast_or_null<DenseElementsAttr>();
auto rhsAttr = operands[1].dyn_cast_or_null<DenseElementsAttr>();

if (lhsTy == resultTy) {
if (rhsAttr && rhsAttr.isSplat() && resultETy.isa<FloatType>()) {
if (rhsAttr.getSplatValue<APFloat>().isZero())
return getInput1();
}
}

if (lhsTy != rhsTy) {
if (lhsAttr && rhsAttr) {
if (lhsTy == resultTy && rhsAttr.isSplat()) {
APFloat r = rhsAttr.getSplatValue<APFloat>();
std::vector<APFloat> v;
v.resize(lhsAttr.size(), APFloat(0.0));
for(int i=0;i<lhsAttr.size(); ++i) {
v[i] = lhsAttr.getValues<APFloat>()[i] + r;
}
return DenseElementsAttr::get(resultTy, v);
}
}
}


if (lhsAttr && lhsAttr.isSplat() && resultETy.isa<FloatType>()) {
if (lhsAttr.getSplatValue<APFloat>().isZero())
return getInput2();
}

if (rhsAttr && rhsAttr.isSplat() && resultETy.isa<FloatType>()) {
if (rhsAttr.getSplatValue<APFloat>().isZero())
return getInput1();
}

if (lhsAttr && lhsAttr.isSplat() && resultETy.isa<IntegerType>()) {
if (lhsAttr.getSplatValue<APInt>().isZero())
return getInput2();
}

if (rhsAttr && rhsAttr.isSplat() && resultETy.isa<IntegerType>()) {
if (rhsAttr.getSplatValue<APInt>().isZero())
return getInput1();
}
if (lhsTy == resultTy && isSplatZero(resultETy, rhsAttr))
return getInput1();
if (rhsTy == resultTy && isSplatZero(resultETy, lhsAttr))
return getInput2();

if (!lhsAttr || !rhsAttr)
return {};

return binaryFolder<std::plus<APInt>, std::plus<APFloat>>(lhsAttr, rhsAttr,
lhsTy);
resultTy);
}

OpFoldResult DivOp::fold(ArrayRef<Attribute> operands) {
Expand Down Expand Up @@ -603,50 +500,26 @@ OpFoldResult MulOp::fold(ArrayRef<Attribute> operands) {
auto resultTy = getType().dyn_cast<RankedTensorType>();
if (!lhsTy || !rhsTy || !resultTy)
return {};
if (lhsTy != rhsTy)
return {};

auto resultETy = resultTy.getElementType();
auto lhsAttr = operands[0].dyn_cast_or_null<DenseElementsAttr>();
auto rhsAttr = operands[1].dyn_cast_or_null<DenseElementsAttr>();

if (lhsAttr && lhsAttr.isSplat() && resultETy.isa<FloatType>()) {
auto val = lhsAttr.getSplatValue<APFloat>();
if (val.isZero())
const int64_t shift = resultETy.isa<IntegerType>() ? getShift() : 0;
if (rhsTy == resultTy) {
if (isSplatZero(resultETy, lhsAttr))
return lhsAttr;
if (val.isExactlyValue(1.0))
if (isSplatOne(resultETy, lhsAttr, shift))
return rhs;
}

if (rhsAttr && rhsAttr.isSplat() && resultETy.isa<FloatType>()) {
auto val = rhsAttr.getSplatValue<APFloat>();
if (val.isZero())
return rhsAttr;
if (val.isExactlyValue(1.0))
return lhs;
}

if (lhsAttr && lhsAttr.isSplat() && resultETy.isa<IntegerType>()) {
auto val = lhsAttr.getSplatValue<APInt>();
if (val.isZero())
return lhsAttr;
const int64_t shift = getShift();
const int64_t shifted = 1LL << shift;
if (val.getSExtValue() == shifted)
return rhs;
}

if (rhsAttr && rhsAttr.isSplat() && resultETy.isa<IntegerType>()) {
auto val = rhsAttr.getSplatValue<APInt>();
const int64_t shift = getShift();
const int64_t shifted = 1LL << shift;
if (val.isZero())
if (lhsTy == resultTy) {
if (isSplatZero(resultETy, rhsAttr))
return rhsAttr;
if (val.getSExtValue() == shifted)
if (isSplatOne(resultETy, rhsAttr, shift))
return lhs;
}

return mulBinaryFolder(lhsAttr, rhsAttr, lhsTy, getShift());
return mulBinaryFolder(lhsAttr, rhsAttr, resultTy, getShift());
}

OpFoldResult SubOp::fold(ArrayRef<Attribute> operands) {
Expand All @@ -655,28 +528,18 @@ OpFoldResult SubOp::fold(ArrayRef<Attribute> operands) {
auto resultTy = getType().dyn_cast<RankedTensorType>();
if (!lhsTy || !rhsTy || !resultTy)
return {};
if (lhsTy != rhsTy)
return {};

auto resultETy = resultTy.getElementType();
auto lhsAttr = operands[0].dyn_cast_or_null<DenseElementsAttr>();
auto rhsAttr = operands[1].dyn_cast_or_null<DenseElementsAttr>();

if (rhsAttr && rhsAttr.isSplat() && resultETy.isa<FloatType>()) {
if (rhsAttr.getSplatValue<APFloat>().isZero())
return getInput1();
}

if (rhsAttr && rhsAttr.isSplat() && resultETy.isa<IntegerType>()) {
if (rhsAttr.getSplatValue<APInt>().isZero())
return getInput1();
}
if (lhsTy == resultTy && isSplatZero(resultETy, rhsAttr))
return getInput1();

if (!lhsAttr || !rhsAttr)
return {};

return binaryFolder<std::minus<APInt>, std::minus<APFloat>>(lhsAttr, rhsAttr,
lhsTy);
resultTy);
}

namespace {
Expand Down Expand Up @@ -917,7 +780,7 @@ OpFoldResult RsqrtOp::fold(FoldAdaptor adaptor) {
auto operand = adaptor.getInput1().dyn_cast_or_null<ElementsAttr>();
if (!operand)
return {};

if (!inputTy.getElementType().isF32())
return {};

Expand Down Expand Up @@ -947,7 +810,7 @@ OpFoldResult PowOp::fold(FoldAdaptor adaptor) {
auto operand2 = adaptor.getInput2().dyn_cast_or_null<ElementsAttr>();
if (!operand2)
return {};

if (!operand1.getElementType().isF32() || !operand2.getElementType().isF32())
return {};

Expand All @@ -961,7 +824,7 @@ OpFoldResult PowOp::fold(FoldAdaptor adaptor) {

OpFoldResult ReciprocalOp::fold(FoldAdaptor adaptor) {
auto src = adaptor.getInput1().dyn_cast_or_null<mlir::DenseElementsAttr>();

if (!src)
return nullptr;

Expand Down Expand Up @@ -989,7 +852,6 @@ OpFoldResult ReverseOp::fold(ArrayRef<Attribute> operands) {
return {};
}


OpFoldResult SliceOp::fold(ArrayRef<Attribute> operands) {
auto inputTy = getInput().getType().dyn_cast<RankedTensorType>();
auto outputTy = getType().dyn_cast<RankedTensorType>();
Expand Down
29 changes: 15 additions & 14 deletions mlir/test/Dialect/Tosa/canonicalize.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,15 +7,15 @@ func.func @argmax_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
return %0 : tensor<?x1xf32>
}

// CHECK-LABEL: @add_zero_different_shape
func.func @add_zero_different_shape(%arg0: tensor<2x3xi32>) -> tensor<4x2x3xi32> {
// CHECK: tosa.add
%zeros = "tosa.const"() {value = dense<0> : tensor<4x2x3xi32>} : () -> tensor<4x2x3xi32>
%1 = "tosa.add"(%arg0, %zeros) : (tensor<2x3xi32>, tensor<4x2x3xi32>) -> tensor<4x2x3xi32>
// CHECK-LABEL: @add_bcast_zero_int
func.func @add_bcast_zero_int(%arg0: tensor<4x2x3xi32>) -> tensor<4x2x3xi32> {
// CHECK-NOT: tosa.add
// CHECK: return %arg0
%zeros = "tosa.const"() {value = dense<0> : tensor<1x1x1xi32>} : () -> tensor<1x1x1xi32>
%1 = "tosa.add"(%arg0, %zeros) : (tensor<4x2x3xi32>, tensor<1x1x1xi32>) -> tensor<4x2x3xi32>
return %1 : tensor<4x2x3xi32>
}


// CHECK-LABEL: @add_zero_int
func.func @add_zero_int(%arg0: tensor<2x3xi32>) -> tensor<2x3xi32> {
// CHECK: return %arg0
Expand Down Expand Up @@ -176,14 +176,6 @@ func.func @pad_determine_val_quant(%arg0: tensor<?x?xi32>, %arg1 : tensor<2x2xi3
return %1 : tensor<?x?xi32>
}

// CHECK-LABEL: @mul_one_different_shape
func.func @mul_one_different_shape(%arg0: tensor<2x3xf32>) -> tensor<4x2x3xf32> {
// CHECK: tosa.mul
%ones = "tosa.const"() {value = dense<1.0> : tensor<4x2x3xf32>} : () -> tensor<4x2x3xf32>
%1 = "tosa.mul"(%arg0, %ones) {shift = 0 : i32} : (tensor<2x3xf32>, tensor<4x2x3xf32>) -> tensor<4x2x3xf32>
return %1 : tensor<4x2x3xf32>
}

// CHECK-LABEL: @mul_one_float
func.func @mul_one_float(%arg0: tensor<2x3xf32>) -> tensor<2x3xf32> {
// CHECK: return %arg0
Expand All @@ -193,6 +185,15 @@ func.func @mul_one_float(%arg0: tensor<2x3xf32>) -> tensor<2x3xf32> {
return %1 : tensor<2x3xf32>
}

// CHECK-LABEL: @mul_bcast_one_float
func.func @mul_bcast_one_float(%arg0: tensor<2x3xf32>) -> tensor<2x3xf32> {
// CHECK: return %arg0
// CHECK-NOT: tosa.mul
%ones = "tosa.const"() {value = dense<1.0> : tensor<1x1xf32>} : () -> tensor<1x1xf32>
%1 = "tosa.mul"(%ones, %arg0) {shift = 0 : i32} : (tensor<1x1xf32>, tensor<2x3xf32>) -> tensor<2x3xf32>
return %1 : tensor<2x3xf32>
}

// CHECK-LABEL: @mul_one_int
func.func @mul_one_int(%arg0: tensor<2x3xi32>) -> tensor<2x3xi32> {
// CHECK: return %arg0
Expand Down