Skip to content

TOSA: fold cast-to-bf16(cast-to-f32(x)) -> cast-to-bf16(x) #413

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
mgehre-amd merged 2 commits into from
Dec 11, 2024
Merged

TOSA: fold cast-to-bf16(cast-to-f32(x)) -> cast-to-bf16(x) #413

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
ff2485b
TOSA: fold cast-to-bf16(cast-to-f32(x)) -> cast-to-bf16(x)
mgehre-amd Dec 10, 2024
729187c
Restrict to bf16-f32-bf16
mgehre-amd Dec 11, 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
12 changes: 12 additions & 0 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 @@ -1008,6 +1008,18 @@ OpFoldResult CastOp::fold(FoldAdaptor adaptor) {
}
}

// Fold cast from bf16 -> f32 -> bf16 into no-op.
if (auto cast = getInput().getDefiningOp<CastOp>()) {
auto sourceElTy = cast.getInput().getType().getElementType();
auto intermediateElTy = cast.getType().getElementType();
auto finalElTy = getType().getElementType();
if (isa<BFloat16Type>(sourceElTy) && isa<Float32Type>(intermediateElTy) &&
isa<BFloat16Type>(finalElTy)) {
getInputMutable().assign(cast.getInput());
return getResult();
}
}

auto operand = llvm::dyn_cast_if_present<ElementsAttr>(adaptor.getInput());
if (!operand)
return {};
Expand Down
8 changes: 8 additions & 0 deletions mlir/test/Dialect/Tosa/canonicalize.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -55,6 +55,14 @@ func.func @cast_fold_double(%arg0: tensor<?x1xf32>) -> tensor<?x1xi8> {
return %1 : tensor<?x1xi8>
}

// CHECK-LABEL: @cast_fold_double
func.func @cast_fold_double2(%arg0: tensor<?x1xbf16>) -> tensor<?x1xbf16> {
// CHECK: return %arg0
%0 = tosa.cast %arg0 : (tensor<?x1xbf16>) -> tensor<?x1xf32>
%1 = tosa.cast %0 : (tensor<?x1xf32>) -> tensor<?x1xbf16>
return %1 : tensor<?x1xbf16>
}

// CHECK-LABEL: @cast_no_fold_double1
func.func @cast_no_fold_double1(%arg0: tensor<?x1xf32>) -> tensor<?x1xi8> {
// CHECK: tosa.cast{{.*}} (tensor<?x1xf32>) -> tensor<?x1xui16>
Expand Down
Loading