https://bugs.llvm.org/show_bug.cgi?id=45748
Bug ID: 45748
Summary: Bad optimization of reverse-initialization of 128-bit
vector with 64-bit vectors
Product: libraries
Version: trunk
Hardware: PC
OS: Linux
Status: NEW
Severity: enhancement
Priority: P
Component: Backend: X86
Assignee: unassignedb...@nondot.org
Reporter: gabrav...@gmail.com
CC: craig.top...@gmail.com, llvm-bugs@lists.llvm.org,
llvm-...@redking.me.uk, spatel+l...@rotateright.com
typedef int32_t v2si __attribute__((vector_size(8)));
typedef int64_t v2di __attribute__((vector_size(16)));
v2di setr_epi64(v2si q0, v2si q1)
{
return v2di{(int64_t)q0, (int64_t)q1};
}
GCC -O3 produces :
setr_epi64(int __vector(2), int __vector(2)):
punpcklqdq xmm0, xmm1
ret
LLVM produces :
setr_epi64(int __vector(2), int __vector(2)): # @setr_epi64(int __vector(2),
int __vector(2))
movq rax, xmm0
movq rcx, xmm1
movq xmm1, rcx
movq xmm0, rax
punpcklqdq xmm0, xmm1 # xmm0 = xmm0[0],xmm1[0]
ret
The code here cannot be further minimized, from what I've been able to see with
my tests.
See also https://godbolt.org/z/ZC-SWs
--
You are receiving this mail because:
You are on the CC list for the bug._______________________________________________
llvm-bugs mailing list
llvm-bugs@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
