Landing: e44dc11c4d9b
Project / Subsystem
gcc / x86
Date
2026-08-19
Author
H.J. Lu
Commit
e44dc11c4d9b80e2dce5a91c0ff07606a3b527b0
Source
github
Perf win
Yes
Breaking
No
All attributes
- project
- gcc
- subsystem
- x86
- patch_id
- —
- commit_hash
- e44dc11c4d9b80e2dce5a91c0ff07606a3b527b0
- source_type
- github
- headline
- x86: Implement truncsfbf2 expansion like vcvtneps2bf16
- tldr
- The default `truncsfbf2` expansion now behaves like `vcvtneps2bf16`, flushing denormals and handling NaNs consistently.
- author
- H.J. Lu
- outcome
- committed
- performance_win
- true
- breaking_change
- false
- series_id
- —
- series_parts
- []
- tags
-
- • x86
- • floating point
- • optimization
- • bugfix
- discussion_id_link
- —
- bugzilla_pr
- —
- date
- 2026-08-19T00:00:00.000Z
The GCC compiler’s expansion for the truncsfbf2 intrinsic on x86 has been updated to mirror the behavior of the vcvtneps2bf16 instruction. This change ensures that input denormal floating-point numbers are flushed to zero, SNANs are quietly converted to QNaNs, and rounding consistently uses the ‘nearest even’ mode. This improves consistency across floating-point conversion operations.