Landing: 1743eb860b73
Project / Subsystem
gcc / middle-end
Date
2026-09-23
Author
Tamar Christina
Commit
1743eb860b739c0bae408536d6a9459f2325fe18
Source
github
Perf win
Yes
Breaking
No
All attributes
- project
- gcc
- subsystem
- middle-end
- patch_id
- —
- commit_hash
- 1743eb860b739c0bae408536d6a9459f2325fe18
- source_type
- github
- headline
- Fold vector unpack operations over polynomial constants.
- tldr
- Compiler optimizes vector unpack operations on variable-length constants for better constant folding.
- author
- Tamar Christina
- outcome
- committed
- performance_win
- true
- breaking_change
- false
- series_id
- —
- series_parts
- []
- tags
-
- • optimization
- • middle-end
- • vectorization
- discussion_id_link
- —
- bugzilla_pr
- —
- date
- 2026-09-23T00:00:00.000Z
The GCC middle-end now folds VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR when operating on suitable variable-length vector constants. This enhancement extends existing folding capabilities for fixed-length vectors to variable-length ones, allowing more constants to be resolved before RTL expansion and potentially eliminating unnecessary instructions.