https://gcc.gnu.org/bugzilla/show_bug.cgi?id=120683
H.J. Lu <hjl.tools at gmail dot com> changed:
What |Removed |Added
----------------------------------------------------------------------------
Summary|vector_loop generates poor |vector_loop/unrolled_loop
|codes on memset/memcpy |generates poor codes on
| |memset/memcpy
--- Comment #3 from H.J. Lu <hjl.tools at gmail dot com> ---
unrolled_loop generates:
[hjl@gnu-tgl-3 pr120683]$ cat x.c
void
foo (char *dest)
{
__builtin_memset (dest, 0, 61);
}
[hjl@gnu-tgl-3 pr120683]$ make x.s
/export/build/gnu/tools-build/gcc-gitlab-debug/build-x86_64-linux/gcc/xgcc
-B/export/build/gnu/tools-build/gcc-gitlab-debug/build-x86_64-linux/gcc/ -O2
-minline-all-stringops
-mmemset-strategy=unrolled_loop:256:noalign,libcall:-1:noalign -mno-sse -S x.c
[hjl@gnu-tgl-3 pr120683]$ cat x.s
.file "x.c"
.text
.p2align 4
.globl foo
.type foo, @function
foo:
.LFB0:
.cfi_startproc
xorl %eax, %eax
.L2:
movl %eax, %edx
movq $0, (%rdi,%rdx)
movq $0, 8(%rdi,%rdx)
movq $0, 16(%rdi,%rdx)
movq $0, 24(%rdi,%rdx)
addl $32, %eax
jc .L2 <<<<<<<< Why is this needed?
movb $0, 28(%rdi,%rax)
andq $0, (%rdi,%rax)
andq $0, 8(%rdi,%rax)
andq $0, 16(%rdi,%rax)
andl $0, 24(%rdi,%rax)
ret
.cfi_endproc
.LFE0:
.size foo, .-foo
.ident "GCC: (GNU) 16.0.0 20250617 (experimental)"
.section .note.GNU-stack,"",@progbits
[hjl@gnu-tgl-3 pr120683]$
