Skip to content

[mlir]Fix compose subview #80551

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 4 commits into from
Feb 7, 2024
Merged

[mlir]Fix compose subview #80551

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
64ad5c6
fix compose-subview.
linuxlonelyeagle Feb 3, 2024
1632e3a
fix nit and update test.
linuxlonelyeagle Feb 6, 2024
ac87e54
Eliminating unnecessary loops.
linuxlonelyeagle Feb 6, 2024
a2a1b96
fix nit.
linuxlonelyeagle Feb 7, 2024
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
89 changes: 51 additions & 38 deletions mlir/lib/Dialect/MemRef/Transforms/ComposeSubView.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,8 +24,7 @@ using namespace mlir;

namespace {

// Replaces a subview of a subview with a single subview. Only supports subview
// ops with static sizes and static strides of 1 (both static and dynamic
// Replaces a subview of a subview with a single subview(both static and dynamic
// offsets are supported).
struct ComposeSubViewOpPattern : public OpRewritePattern<memref::SubViewOp> {
using OpRewritePattern::OpRewritePattern;
Expand All @@ -51,64 +50,78 @@ struct ComposeSubViewOpPattern : public OpRewritePattern<memref::SubViewOp> {
}

// Offsets, sizes and strides OpFoldResult for the combined 'SubViewOp'.
SmallVector<OpFoldResult> offsets, sizes, strides;

// Because we only support input strides of 1, the output stride is also
// always 1.
if (llvm::all_of(strides, [](OpFoldResult &valueOrAttr) {
Attribute attr = llvm::dyn_cast_if_present<Attribute>(valueOrAttr);
return attr && cast<IntegerAttr>(attr).getInt() == 1;
})) {
strides = SmallVector<OpFoldResult>(sourceOp.getMixedStrides().size(),
rewriter.getI64IntegerAttr(1));
} else {
return failure();
SmallVector<OpFoldResult> offsets, sizes, strides,
opStrides = op.getMixedStrides(),
sourceStrides = sourceOp.getMixedStrides();

// The output stride in each dimension is equal to the product of the
// dimensions corresponding to source and op.
int64_t sourceStrideValue;
for (auto &&[opStride, sourceStride] :
llvm::zip(opStrides, sourceStrides)) {
Attribute opStrideAttr = dyn_cast_if_present<Attribute>(opStride);
Attribute sourceStrideAttr = dyn_cast_if_present<Attribute>(sourceStride);
if (!opStrideAttr || !sourceStrideAttr)
return failure();
sourceStrideValue = cast<IntegerAttr>(sourceStrideAttr).getInt();
strides.push_back(rewriter.getI64IntegerAttr(
cast<IntegerAttr>(opStrideAttr).getInt() * sourceStrideValue));
}

// The rules for calculating the new offsets and sizes are:
// * Multiple subview offsets for a given dimension compose additively.
// ("Offset by m" followed by "Offset by n" == "Offset by m + n")
// ("Offset by m and Stride by k" followed by "Offset by n" == "Offset by
// m + n * k")
// * Multiple sizes for a given dimension compose by taking the size of the
// final subview and ignoring the rest. ("Take m values" followed by "Take
// n values" == "Take n values") This size must also be the smallest one
// by definition (a subview needs to be the same size as or smaller than
// its source along each dimension; presumably subviews that are larger
// than their sources are disallowed by validation).
for (auto it : llvm::zip(op.getMixedOffsets(), sourceOp.getMixedOffsets(),
op.getMixedSizes())) {
auto opOffset = std::get<0>(it);
auto sourceOffset = std::get<1>(it);
auto opSize = std::get<2>(it);

for (auto &&[opOffset, sourceOffset, sourceStride, opSize] :
llvm::zip(op.getMixedOffsets(), sourceOp.getMixedOffsets(),
sourceOp.getMixedStrides(), op.getMixedSizes())) {
// We only support static sizes.
if (opSize.is<Value>()) {
return failure();
}

sizes.push_back(opSize);
Attribute opOffsetAttr = llvm::dyn_cast_if_present<Attribute>(opOffset),
sourceOffsetAttr =
llvm::dyn_cast_if_present<Attribute>(sourceOffset);

llvm::dyn_cast_if_present<Attribute>(sourceOffset),
sourceStrideAttr =
llvm::dyn_cast_if_present<Attribute>(sourceStride);
if (opOffsetAttr && sourceOffsetAttr) {

// If both offsets are static we can simply calculate the combined
// offset statically.
offsets.push_back(rewriter.getI64IntegerAttr(
cast<IntegerAttr>(opOffsetAttr).getInt() +
cast<IntegerAttr>(opOffsetAttr).getInt() *
cast<IntegerAttr>(sourceStrideAttr).getInt() +
cast<IntegerAttr>(sourceOffsetAttr).getInt()));
} else {
// When either offset is dynamic, we must emit an additional affine
// transformation to add the two offsets together dynamically.
AffineExpr expr = rewriter.getAffineConstantExpr(0);
AffineExpr expr;
SmallVector<Value> affineApplyOperands;
for (auto valueOrAttr : {opOffset, sourceOffset}) {
if (auto attr = llvm::dyn_cast_if_present<Attribute>(valueOrAttr)) {
expr = expr + cast<IntegerAttr>(attr).getInt();
} else {
expr =
expr + rewriter.getAffineSymbolExpr(affineApplyOperands.size());
affineApplyOperands.push_back(valueOrAttr.get<Value>());
}

// Make 'expr' add 'sourceOffset'.
if (auto attr = llvm::dyn_cast_if_present<Attribute>(sourceOffset)) {
expr =
rewriter.getAffineConstantExpr(cast<IntegerAttr>(attr).getInt());
} else {
expr = rewriter.getAffineSymbolExpr(affineApplyOperands.size());
affineApplyOperands.push_back(sourceOffset.get<Value>());
}

// Multiply 'opOffset' by 'sourceStride' and make the 'expr' add the
// result.
if (auto attr = llvm::dyn_cast_if_present<Attribute>(opOffset)) {
expr = expr + cast<IntegerAttr>(attr).getInt() *
cast<IntegerAttr>(sourceStrideAttr).getInt();
} else {
expr =
expr + rewriter.getAffineSymbolExpr(affineApplyOperands.size()) *
cast<IntegerAttr>(sourceStrideAttr).getInt();
affineApplyOperands.push_back(opOffset.get<Value>());
}

AffineMap map = AffineMap::get(0, affineApplyOperands.size(), expr);
Expand All @@ -120,8 +133,8 @@ struct ComposeSubViewOpPattern : public OpRewritePattern<memref::SubViewOp> {

// This replaces 'op' but leaves 'sourceOp' alone; if it no longer has any
// uses it can be removed by a (separate) dead code elimination pass.
rewriter.replaceOpWithNewOp<memref::SubViewOp>(op, sourceOp.getSource(),
offsets, sizes, strides);
rewriter.replaceOpWithNewOp<memref::SubViewOp>(
op, op.getType(), sourceOp.getSource(), offsets, sizes, strides);
return success();
}
};
Expand Down
86 changes: 71 additions & 15 deletions mlir/test/Transforms/compose-subview.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,18 +1,20 @@
// RUN: mlir-opt %s -test-compose-subview -split-input-file | FileCheck %s

func.func @main(%input: memref<4x1024xf32>) -> memref<1x128xf32, strided<[1024, 1], offset: 3456>> {
// CHECK: subview %arg0[3, 384] [1, 128] [1, 1]
// CHECK-SAME: memref<4x1024xf32> to memref<1x128xf32, strided<[1024, 1], offset: 3456>>
// CHECK-LABEL: func.func @subview_strided(
// CHECK-SAME: %[[VAL_0:.*]]: memref<4x1024xf32>) -> memref<1x128xf32, strided<[1024, 1], offset: 3456>> {
func.func @subview_strided(%input: memref<4x1024xf32>) -> memref<1x128xf32, strided<[1024, 1], offset: 3456>> {
// CHECK: %[[VAL_1:.*]] = memref.subview %[[VAL_0]][3, 384] [1, 128] [1, 1] : memref<4x1024xf32> to memref<1x128xf32, strided<[1024, 1], offset: 3456>>
%0 = memref.subview %input[2, 256] [2, 256] [1, 1] : memref<4x1024xf32> to memref<2x256xf32, strided<[1024, 1], offset: 2304>>
%1 = memref.subview %0[1, 128] [1, 128] [1, 1] : memref<2x256xf32, strided<[1024, 1], offset: 2304>> to memref<1x128xf32, strided<[1024, 1], offset: 3456>>
return %1 : memref<1x128xf32, strided<[1024, 1], offset: 3456>>
}

// -----

func.func @main(%input: memref<4x1024xf32>) -> memref<1x10xf32, strided<[1024, 1], offset: 3745>> {
// CHECK: subview %arg0[3, 673] [1, 10] [1, 1]
// CHECK-SAME: memref<4x1024xf32> to memref<1x10xf32, strided<[1024, 1], offset: 3745>>
// CHECK-LABEL: func.func @subview_strided(
// CHECK-SAME: %[[VAL_0:.*]]: memref<4x1024xf32>) -> memref<1x10xf32, strided<[1024, 1], offset: 3745>> {
func.func @subview_strided(%input: memref<4x1024xf32>) -> memref<1x10xf32, strided<[1024, 1], offset: 3745>> {
// CHECK: %[[VAL_1:.*]] = memref.subview %[[VAL_0]][3, 673] [1, 10] [1, 1] : memref<4x1024xf32> to memref<1x10xf32, strided<[1024, 1], offset: 3745>>
%0 = memref.subview %input[1, 512] [3, 256] [1, 1] : memref<4x1024xf32> to memref<3x256xf32, strided<[1024, 1], offset: 1536>>
%1 = memref.subview %0[1, 128] [2, 128] [1, 1] : memref<3x256xf32, strided<[1024, 1], offset: 1536>> to memref<2x128xf32, strided<[1024, 1], offset: 2688>>
%2 = memref.subview %1[1, 33] [1, 10] [1, 1] : memref<2x128xf32, strided<[1024, 1], offset: 2688>> to memref<1x10xf32, strided<[1024, 1], offset: 3745>>
Expand All @@ -21,27 +23,81 @@ func.func @main(%input: memref<4x1024xf32>) -> memref<1x10xf32, strided<[1024, 1

// -----

func.func @main(%input: memref<4x1024xf32>) -> memref<1x128xf32, strided<[1024, 1], offset: ?>> {
// CHECK: [[CST_3:%.*]] = arith.constant 3 : index
// CHECK-LABEL: func.func @subview_strided(
// CHECK-SAME: %[[VAL_0:.*]]: memref<4x1024xf32>) -> memref<1x128xf32, strided<[1024, 1], offset: ?>> {
func.func @subview_strided(%input: memref<4x1024xf32>) -> memref<1x128xf32, strided<[1024, 1], offset: ?>> {
// CHECK: %[[VAL_1:.*]] = arith.constant 3 : index
%cst_1 = arith.constant 1 : index
%cst_2 = arith.constant 2 : index
// CHECK: subview %arg0{{\[}}[[CST_3]], 384] [1, 128] [1, 1]
// CHECK-SAME: memref<4x1024xf32> to memref<1x128xf32, strided<[1024, 1], offset: ?>>
// CHECK: %[[VAL_2:.*]] = memref.subview %[[VAL_0]]{{\[}}%[[VAL_1]], 384] [1, 128] [1, 1] : memref<4x1024xf32> to memref<1x128xf32, strided<[1024, 1], offset: ?>>
%0 = memref.subview %input[%cst_2, 256] [2, 256] [1, 1] : memref<4x1024xf32> to memref<2x256xf32, strided<[1024, 1], offset: ?>>
%1 = memref.subview %0[%cst_1, 128] [1, 128] [1, 1] : memref<2x256xf32, strided<[1024, 1], offset: ?>> to memref<1x128xf32, strided<[1024, 1], offset: ?>>
return %1 : memref<1x128xf32, strided<[1024, 1], offset: ?>>
}

// -----

func.func @main(%input: memref<4x1024xf32>) -> memref<1x128xf32, strided<[1024, 1], offset: ?>> {
// CHECK: [[CST_3:%.*]] = arith.constant 3 : index
// CHECK-LABEL: func.func @subview_strided(
// CHECK-SAME: %[[VAL_0:.*]]: memref<4x1024xf32>) -> memref<1x128xf32, strided<[1024, 1], offset: ?>> {
func.func @subview_strided(%input: memref<4x1024xf32>) -> memref<1x128xf32, strided<[1024, 1], offset: ?>> {
// CHECK: %[[VAL_1:.*]] = arith.constant 3 : index
%cst_2 = arith.constant 2 : index
// CHECK: [[CST_384:%.*]] = arith.constant 384 : index
// CHECK: %[[VAL_2:.*]] = arith.constant 384 : index
%cst_128 = arith.constant 128 : index
// CHECK: subview %arg0{{\[}}[[CST_3]], [[CST_384]]] [1, 128] [1, 1]
// CHECK-SAME: memref<4x1024xf32> to memref<1x128xf32, strided<[1024, 1], offset: ?>>
// CHECK: %[[VAL_3:.*]] = memref.subview %[[VAL_0]]{{\[}}%[[VAL_1]], %[[VAL_2]]] [1, 128] [1, 1] : memref<4x1024xf32> to memref<1x128xf32, strided<[1024, 1], offset: ?>>
%0 = memref.subview %input[%cst_2, 256] [2, 256] [1, 1] : memref<4x1024xf32> to memref<2x256xf32, strided<[1024, 1], offset: ?>>
%1 = memref.subview %0[1, %cst_128] [1, 128] [1, 1] : memref<2x256xf32, strided<[1024, 1], offset: ?>> to memref<1x128xf32, strided<[1024, 1], offset: ?>>
return %1 : memref<1x128xf32, strided<[1024, 1], offset: ?>>
}

// -----

// CHECK-LABEL: func.func @subview_strided(
// CHECK-SAME: %[[VAL_0:.*]]: memref<4x1024xf32>) -> memref<1x64xf32, strided<[4096, 4], offset: 4480>> {
func.func @subview_strided(%input: memref<4x1024xf32>) -> memref<1x64xf32, strided<[4096, 4], offset: 4480>> {
// CHECK: %[[VAL_1:.*]] = memref.subview %[[VAL_0]][4, 384] [1, 64] [4, 4] : memref<4x1024xf32> to memref<1x64xf32, strided<[4096, 4], offset: 4480>>
%0 = memref.subview %input[2, 256] [2, 256] [2, 2] : memref<4x1024xf32> to memref<2x256xf32, strided<[2048, 2], offset: 2304>>
%1 = memref.subview %0[1, 64] [1, 64] [2, 2] : memref<2x256xf32, strided<[2048, 2], offset: 2304>> to memref<1x64xf32, strided<[4096, 4], offset: 4480>>
return %1 : memref<1x64xf32, strided<[4096, 4], offset: 4480>>
}

// -----

// CHECK-LABEL: func.func @subview_strided(
// CHECK-SAME: %[[VAL_0:.*]]: memref<30x30xf32>) -> memref<2x2xf32, strided<[240, 8], offset: 217>> {
func.func @subview_strided(%input: memref<30x30xf32>) -> memref<2x2xf32, strided<[240, 8], offset: 217>> {
// CHECK: %[[VAL_1:.*]] = memref.subview %[[VAL_0]][7, 7] [2, 2] [8, 8] : memref<30x30xf32> to memref<2x2xf32, strided<[240, 8], offset: 217>>
%0 = memref.subview %input[1, 1] [12, 12] [2, 2] : memref<30x30xf32> to memref<12x12xf32, strided<[60, 2], offset: 31>>
%1 = memref.subview %0[1, 1] [5, 5] [2, 2] : memref<12x12xf32, strided<[60, 2], offset: 31>> to memref<5x5xf32, strided<[120, 4], offset: 93>>
%2 = memref.subview %1[1, 1] [2, 2] [2, 2] : memref<5x5xf32, strided<[120, 4], offset: 93>> to memref<2x2xf32, strided<[240, 8], offset: 217>>
return %2 : memref<2x2xf32, strided<[240, 8], offset: 217>>
}

// -----

// CHECK-LABEL: func.func @subview_strided(
// CHECK-SAME: %[[VAL_0:.*]]: memref<4x1024xf32>) -> memref<1x64xf32, strided<[4096, 4], offset: ?>> {
func.func @subview_strided(%input: memref<4x1024xf32>) -> memref<1x64xf32, strided<[4096, 4], offset: ?>> {
// CHECK: %[[VAL_1:.*]] = arith.constant 4 : index
%cst_2 = arith.constant 2 : index
// CHECK: %[[VAL_2:.*]] = arith.constant 384 : index
%cst_64 = arith.constant 64 : index
// CHECK: %[[VAL_3:.*]] = memref.subview %[[VAL_0]]{{\[}}%[[VAL_1]], %[[VAL_2]]] [1, 64] [4, 4] : memref<4x1024xf32> to memref<1x64xf32, strided<[4096, 4], offset: ?>>
%0 = memref.subview %input[%cst_2, 256] [2, 256] [2, 2] : memref<4x1024xf32> to memref<2x256xf32, strided<[2048, 2], offset: ?>>
%1 = memref.subview %0[1, %cst_64] [1, 64] [2, 2] : memref<2x256xf32, strided<[2048, 2], offset: ?>> to memref<1x64xf32, strided<[4096, 4], offset: ?>>
return %1 : memref<1x64xf32, strided<[4096, 4], offset: ?>>
}

// -----

// CHECK-LABEL: func.func @subview_strided(
// CHECK-SAME: %[[VAL_0:.*]]: memref<4x1024xf32>) -> memref<1x64xf32, strided<[4096, 4], offset: ?>> {
func.func @subview_strided(%input: memref<4x1024xf32>) -> memref<1x64xf32, strided<[4096, 4], offset: ?>> {
// CHECK: %[[VAL_1:.*]] = arith.constant 4 : index
%cst_1 = arith.constant 1 : index
%cst_2 = arith.constant 2 : index
// CHECK: %[[VAL_2:.*]] = memref.subview %[[VAL_0]]{{\[}}%[[VAL_1]], 384] [1, 64] [4, 4] : memref<4x1024xf32> to memref<1x64xf32, strided<[4096, 4], offset: ?>>
%0 = memref.subview %input[%cst_2, 256] [2, 256] [2, 2] : memref<4x1024xf32> to memref<2x256xf32, strided<[2048, 2], offset: ?>>
%1 = memref.subview %0[%cst_1, 64] [1, 64] [2, 2] : memref<2x256xf32, strided<[2048, 2], offset: ?>> to memref<1x64xf32, strided<[4096, 4], offset: ?>>
return %1 : memref<1x64xf32, strided<[4096, 4], offset: ?>>
}