Landing: 4024e780b148

Project / Subsystem

gcc / amdgcn

Date

2026-09-21

Author

Andrew Stubbs

Commit

4024e780b14853d614dfd9ed002b04fc88cb75f9

Source

github

Perf win

No

Breaking

No

All attributes

project
gcc
subsystem
amdgcn
patch_id
—
commit_hash
4024e780b14853d614dfd9ed002b04fc88cb75f9
source_type
github
headline
Fix vector popcount instruction for AMD GCN targets
tldr
Corrects the vector version of the popcount instruction on AMD GCN architectures.
author
Andrew Stubbs
outcome
committed
performance_win
false
breaking_change
false
series_id
—
series_parts
[]
tags
  • • amdgcn
  • • gpu
  • • optimization
discussion_id_link
—
bugzilla_pr
—
date
2026-09-21T00:00:00.000Z

A bug in the vector version of the population count (popcount) instruction for AMD GCN targets has been fixed. The v_bcnt instruction was missing its third parameter, which is present in the scalar version. This commit adds the necessary parameter, ensuring correct execution of vector popcount operations.