Skip to content
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

Stop forcing approximation of math.erf with --iree-codegen-gpu-native-math-precision #20074

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
+0 −7
Open

Stop forcing approximation of math.erf with --iree-codegen-gpu-native-math-precision #20074

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
7 changes: 0 additions & 7 deletions compiler/src/iree/compiler/Codegen/Common/MathTransformPass.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -100,13 +100,6 @@ static bool predicateF32Cast(StringRef name,
static bool predicateApprox(StringRef name,
IREE::HAL::ExecutableTargetAttr target) {
if (clNativeMathPrecision) { // Legacy.
if (name == math::ErfOp::getOperationName()) {
// The legacy implementation had a bug: it always applied polynomial
// approximation of math.erf, even when clNativeMathPrecision was passed.
// We actually have CI tests that rely on that bug: they pass
// clNativeMathPrecision but fail unless math.erf is approximated.
return true;
}
return false;
}
StringRef acos = math::AcosOp::getOperationName();
Expand Down
Loading