Skip to content

JIT: better addressing mode for floating point on arm64 #65468

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 2 commits into from
Feb 17, 2022
Merged

JIT: better addressing mode for floating point on arm64 #65468

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
3f17351
Add addressing mode support for floats
EgorBo Feb 16, 2022
e3032a0
Attempt to fix regressions
EgorBo Feb 17, 2022
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: 1 addition & 1 deletion src/coreclr/jit/lower.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5367,7 +5367,7 @@ bool Lowering::TryCreateAddrMode(GenTree* addr, bool isContainable, GenTree* par
#ifdef TARGET_ARM64
// Check if we can "contain" LEA(BFIZ) in order to extend 32bit index to 64bit as part of load/store.
if ((index != nullptr) && index->OperIs(GT_BFIZ) && index->gtGetOp1()->OperIs(GT_CAST) &&
index->gtGetOp2()->IsCnsIntOrI() && varTypeIsIntegral(targetType))
index->gtGetOp2()->IsCnsIntOrI() && (varTypeIsIntegral(targetType) || varTypeIsFloating(targetType)))
{
// BFIZ node is a binary op where op1 is GT_CAST and op2 is GT_CNS_INT
GenTreeCast* cast = index->gtGetOp1()->AsCast();
Expand Down
2 changes: 1 addition & 1 deletion src/coreclr/jit/morph.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5511,7 +5511,7 @@ GenTree* Compiler::fgMorphArrayIndex(GenTree* tree)
// See https://github.com/dotnet/runtime/pull/61293#issuecomment-964146497

// Use 2) form only for primitive types for now - it significantly reduced number of size regressions
if (!varTypeIsIntegral(elemTyp))
if (!varTypeIsIntegral(elemTyp) && !varTypeIsFloating(elemTyp))
{
groupArrayRefWithElemOffset = false;
}
Expand Down