https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115518
Bug ID: 115518
Summary: aarch64: Poor codegen for arm_neon_sve_bridge.h
Product: gcc
Version: 15.0
Status: UNCONFIRMED
Keywords: aarch64-sve
Severity: normal
Priority: P3
Component: target
Assignee: unassigned at gcc dot gnu.org
Reporter: rsandifo at gcc dot gnu.org
Target Milestone: ---
Target: aarch64*-*-*
With PR115464 fixed, the following testcase:
#include <arm_neon.h>
#include <arm_sve.h>
#include <arm_neon_sve_bridge.h>
svuint16_t
convolve4_4_x (uint16x8x2_t permute_tbl, svuint16_t a)
{
return svset_neonq_u16 (a, permute_tbl.val[1]);
}
generates:
mov v0.16b, v1.16b
ptrue p3.h, vl8
sel z0.h, p3, z0.h, z2.h
ret
The move is redundant: we should be able to use z1.h as input to the sel
instead.