llvm-project.git/llvm/lib/Transforms/Vectorize/VectorCombine.cpp, branch main

[VectorCombine] Try to scalarize vector loads feeding bitcast instructions. (#164682)

2025-11-12T15:35:03+00:00

This change aims to convert vector loads to scalar loads, if they are
only converted to scalars after anyway.

alive2 proof: https://alive2.llvm.org/ce/z/U_rvht

[VectorCombine] support mismatching extract/insert indices for foldInsExtFNeg (#126408)

2025-11-07T18:35:40+00:00

insertelt DestVec, (fneg (extractelt SrcVec, Index)), Index 
-> shuffle DestVec, (shuffle (fneg SrcVec), poison, SrcMask), Mask

In previous, the above transform was only possible if the Extract/Insert
Index was the same; this patch makes the above transform possible
even if the two indexes are different.

Proof: https://alive2.llvm.org/ce/z/aDfdyG
Fixes: https://github.com/llvm/llvm-project/issues/125675

[VectorCombine] Avoid inserting freeze when scalarizing extend-extract if all extracts would lead to UB on poison. (#164683)

2025-11-04T12:39:04+00:00

This change aims to avoid inserting a freeze instruction between the
load and bitcast when scalarizing extend-extract. This is particularly
useful in combination with
https://github.com/llvm/llvm-project/pull/164682, which can then
potentially further scalarize, provided there is no freeze.

alive2 proof: https://alive2.llvm.org/ce/z/W-GD88

[VectorCombine] Preserve cast flags in foldBitOpOfCastConstant (#161237)

2025-09-30T08:38:03+00:00

Follow-up of #157822.

[VectorCombine] foldShuffleOfCastops - handle unary shuffles (#160009)

2025-09-29T13:21:59+00:00

Fixes #156853.

[VectorCombine] Fix rotation in phi narrowing. (#160465)

2025-09-29T12:26:35+00:00

Fix bug in #140188 where incoming vectors are rotated in the wrong
direction.

Co-authored-by: Leon Clark

[VectorCombine] Fix scalarizeExtExtract for big-endian (#157962)

2025-09-15T10:08:16+00:00

The scalarizeExtExtract transform assumed little-endian lane ordering,
causing miscompiles on big-endian targets such as AIX/PowerPC under -O3
-flto.

This patch updates the shift calculation to handle endianness correctly
for big-endian targets. No functional change
for little-endian targets.

Fixes #158197.

---------

Co-authored-by: Simon Pilgrim

[VectorCombine] Add Ext and Trunc support in foldBitOpOfCastConstant (#157822)

2025-09-11T09:08:47+00:00

Follow-up of https://github.com/llvm/llvm-project/pull/155216.
This patch doesn't preserve the flags. I will implement it in the
follow-up patch.

[InstCombine][VectorCombine][NFC] Unify uses of lossless inverse cast (#156597)

2025-09-08T13:30:06+00:00

This patch addresses
https://github.com/llvm/llvm-project/pull/155216#discussion_r2297724663.
This patch adds a helper function to put the inverse cast on constants,
with cast flags preserved(optional).
Follow-up patches will add trunc/ext handling on VectorCombine and flags
preservation on InstCombine.

[VectorCombine] foldSelectShuffle - early-out cases where the max vector register width isn't large enough (#157430)

2025-09-08T12:04:23+00:00

Technically this could happen with vector units that can't handle all legal scalar widths - but its good enough to use a generic crash test without a suitable target

Fixes #157335