Landing: f09bafa778ca

← All landings

Project / Subsystem

gcc / libstdc++

Date

2026-07-02

Author

Matthias Kretz

Commit

f09bafa778caaeac1faaa16eacf4d0ad7adb6c92

Source

github

Perf win

Breaking

View on GitHub →

All attributes

project: gcc
subsystem: libstdc++
patch_id: —
commit_hash: f09bafa778caaeac1faaa16eacf4d0ad7adb6c92
source_type: github
headline: Remove x86-specific optimizations from SIMD reduce implementation
tldr: Unnecessary x86-specific code removed from libstdc++'s SIMD reduce implementation.
author: Matthias Kretz
outcome: committed
performance_win: false
breaking_change: false
series_id: —
series_parts: []
tags: • libstdc++
• simd
• optimization
• reduce
discussion_id_link: —
bugzilla_pr: —
url: https://github.com/gcc-mirror/gcc/commit/f09bafa778caaeac1faaa16eacf4d0ad7adb6c92
date: 2026-07-02T00:00:00.000Z

This commit removes an x86-specific implementation of _M_elements_shifted_to_front from libstdc++‘s SIMD reduce functionality. The optimization was previously deemed necessary for efficient reductions but is no longer required, as the generic _S_static_permute implementation now optimizes equally well or better. A separate change tidies up an incorrect macro name in a test file.