Landing: 1782fdf86886
Project / Subsystem
gcc / gcc/i386
Date
2026-07-19
Author
Dipesh Sharma
Commit
1782fdf868869178911504131f512a3982a830ff
Source
github
Perf win
Yes
Breaking
No
All attributes
- project
- gcc
- subsystem
- gcc/i386
- patch_id
- —
- commit_hash
- 1782fdf868869178911504131f512a3982a830ff
- source_type
- github
- headline
- Add AVX10/v2 FP8 to FP32 and vice-versa conversions.
- tldr
- Enables conversions between FP8 and FP32 floating-point formats using AVX10/v2 intrinsics.
- author
- Dipesh Sharma
- outcome
- committed
- performance_win
- true
- breaking_change
- false
- series_id
- —
- series_parts
- []
- tags
-
- • x86
- • AVX
- • AI
- • floating-point
- discussion_id_link
- —
- bugzilla_pr
- —
- date
- 2026-07-19T00:00:00.000Z
This commit introduces a comprehensive set of AVX10/v2 intrinsic functions for converting between FP8 (8-bit floating-point) and FP32 (single-precision floating-point) formats. It adds support for various vector sizes (128, 256, and 512 bits) and includes masked operations. New intrinsic definitions, types, and test cases are provided to ensure the correct implementation of these conversions for AI and high-performance computing workloads.