onnx-mlir/src/Transform/ONNX/Rewrite.td

205 lines
6.6 KiB
TableGen

//===---- ONNXRewrite.td - Pattern Match Rewriting for ONNX --*- tablegen -===//
//
// Copyright 2019 The IBM Research Authors.
//
// =============================================================================
//
// Defines language-specific pattern match optimizations for ONNX using
// Declarative Rewrite Rules (DRR) specified using TableGen records.
//
//===----------------------------------------------------------------------===//
#ifndef ONNX_REWRITE
#define ONNX_REWRITE
#ifndef OP_BASE
include "src/Dialect/ONNX/ONNXOps.td"
#endif // OP_BASE
/// Note: The DRR definition used for defining patterns is shown below:
///
/// class Pattern<
/// dag sourcePattern, list<dag> resultPatterns,
/// list<dag> additionalConstraints = [],
/// dag benefitsAdded = (addBenefit 0)
/// >;
// Create a DenseElementsAttr from a float attribute and an element type.
def createDenseElementsAttrFromFloatAttr : NativeCodeCall<
"createDenseElementsAttrFromFloatAttr($_builder, $0.getType().cast<ShapedType>().getElementType(), $1)">;
// Create a DenseElementsAttr from the shape of the type of a value.
def createDenseElementsAttrFromShape : NativeCodeCall<
"createDenseElementsAttrFromShape($_builder, $0)">;
// Create a DenseElementsAttr from the size of the type of a value.
def createDenseElementsAttrFromSize : NativeCodeCall<
"createDenseElementsAttrFromSize($_builder, $0)">;
// If '$1' is not NoneType, do subtraction '$1 - $2'.
// Otherwise, take the negative of '$2'.
def subtractOrNeg: NativeCodeCall<
"subtractOrNeg($_builder, $0.getDefiningOp()->getLoc(), $1, $2)">;
// Create an ArrayAttr of IntergerAttr(s) of values in [1, N].
def createArrayAttrOfOneToRankOf : NativeCodeCall<
"createArrayAttrOfOneToN($_builder, $0.getType().cast<ShapedType>().getRank() - 1)">;
def GetNullAttr :
NativeCodeCall<"Attribute()">;
// Create a StringAttr from a string.
class StringAttrOfValue<string val>:
NativeCodeCall<"$_builder.getStringAttr(\"" # val # "\")">;
// Create a DenseElementsAttr from an interger value.
// It seems Table-gen does not support `float` type, so we can not pass a float value.
class FloatAttrOfValue<int val>:
NativeCodeCall<"createDenseFloatAttrOfValue($_builder, $0, " # val # ")">;
// Create an ArrayAttr of IntergerAttr(s) of zero values.
// This function is used for padding attribute in Conv.
def createArrayAttrOfZerosFrom:
NativeCodeCall<"createArrayAttrOfZeros($_builder, $0)">;
// Pad a ArrayAttr with zeros.
//
// pads = [B1, B2, ... Bk, E1, E2, ..., Ek]
//
// becomes:
//
// pads = [0,... 0, B1, B2, ... Bk, 0,... 0, E1, E2, ..., Ek]
// |_____| |_____|
// nZeros nZeros
//
// This function is used for padding attribute in Conv.
class insertZerosForNonPaddedDims<int extensionLength>:
NativeCodeCall<"insertZerosForNonPaddedDims($_builder, $0,"
# extensionLength # ")">;
// Check whether an ArrayAttr contains non-zero values or not.
def HasNonZeroInArrayAttr: Constraint<CPred<"hasNonZeroInArrayAttr($_self)">,
"has non-zero elements">;
// Check that a StrAttr does not contain a specific value.
class IsNotStringAttrOfValue<string val>:
Constraint<CPred<"$0.cast<StringAttr>().getValue() != \"" # val # "\"">>;
//===----------------------------------------------------------------------===//
// Rewrite:
// %0 = onnx.ConvOp(%D : tensor<DShape>, %K)
// {pads = [b0, b1, ... bK, e0, e1, ..., eK]} ->
// tensor<OutShape>
//
// as:
// %0 = onnx.PadConstantValuePadOp(%D)
// {pads = [0, 0, b0, b1, ... bK, 0, 0, e0, e1, ..., eK]} ->
// tensor<DPaddedShape>
// %1 = onnx.Conv(%0 : tensor<DPaddedShape>, %K) {pads = [0, ..., 0]} ->
// tensor<OutShape>
//===----------------------------------------------------------------------===//
def ConvOpPaddingPattern: Pat<
(ONNXConvOp:$res
$x,
$w, $b, $auto_pad, $dilation, $group, $kernel_shape,
$pads,
$strides),
(ONNXConvOp
(ONNXPadOp $x,
(ONNXConstantOp (GetNullAttr),
(insertZerosForNonPaddedDims<2> $pads)),
(ONNXConstantOp (GetNullAttr),
(FloatAttrOfValue<0> $res)),
(StringAttrOfValue<"constant">)),
$w, $b, $auto_pad, $dilation, $group, $kernel_shape,
(createArrayAttrOfZerosFrom $pads),
$strides),
[(HasNonZeroInArrayAttr:$pads), (IsNotStringAttrOfValue<"VALID"> $auto_pad)]
>;
//===----------------------------------------------------------------------===//
// This is to fuse the composition: 'BatchNorm o Conv' into 'Conv'
// by deriving new 'w' and 'b' for 'Conv':
//
// We have:
// (Conv) z = w * x + b
// (BatchNorm) y = scale * (z - mean) / sqrt(var + eps) + bias
//
// which corresponds to the following computation:
// y = w_ * x + b_
// where
// w_ = scale * w / sqrt(var + eps)
// b_ = B + scale * (b - mean) / sqrt(var + eps)
//
// Hence, we rewrite:
// onnx.BatchNormalizationTestMode(
// onnx.Conv(x, w, b),
// scale, B, mean, var
// ) {eps = ...}
//
// as:
// onnx.Conv(x, w_, b_)
//
// where
// w_ = scale * w / sqrt(var + eps)
// b_ = B + scale * (b - mean) / sqrt(var + eps)
//
//===----------------------------------------------------------------------===//
def FuseBatchNormTestModeConvPattern: Pat<
(ONNXBatchNormalizationTestModeOp:$res
(ONNXConvOp $x, $w, $b,
$auto_pad, $dilation, $group, $kernel_shape, $pads, $strides),
$scale, $B, $mean, $var, $epsilon, $momentum),
(ONNXConvOp
$x,
// w_
(ONNXMulOp
$w,
(ONNXUnsqueezeOp
(ONNXDivOp:$coefficientW
$scale,
(ONNXSqrtOp
(ONNXAddOp
$var,
(ONNXConstantOp
(GetNullAttr),
(createDenseElementsAttrFromFloatAttr $res, $epsilon))))),
(createArrayAttrOfOneToRankOf $w))),
// b_
(ONNXAddOp
$B,
(ONNXMulOp
$coefficientW,
(subtractOrNeg $res, $b, $mean))),
$auto_pad, $dilation, $group, $kernel_shape, $pads, $strides)
>;
def IsStaticShapeTensor:
Constraint<
CPred<
"$_self.getType().cast<::mlir::ShapedType>().hasStaticShape()">,
"hasStaticShape">;
def ShapeToConstantPattern: Pat<
(ONNXShapeOp $A),
(ONNXConstantOp
(GetNullAttr),
(createDenseElementsAttrFromShape $A)),
[(IsStaticShapeTensor:$A)]
>;
def SizeToConstantPattern: Pat<
(ONNXSizeOp $A),
(ONNXConstantOp
(GetNullAttr),
(createDenseElementsAttrFromSize $A)),
[(IsStaticShapeTensor:$A)]
>;
#endif // ONNX_REWRITE