https://bugs.llvm.org/show_bug.cgi?id=32086
Bug ID: 32086
Summary: Multiple moves compiled inefficiently
Product: new-bugs
Version: trunk
Hardware: PC
OS: Linux
Status: NEW
Severity: enhancement
Priority: P
Component: new bugs
Assignee: [email protected]
Reporter: [email protected]
CC: [email protected]
Consider:
double a[4], b[4], c[4];
void foo ()
{
a[0] = c[0];
a[1] = c[1];
a[2] = c[0];
a[3] = c[1];
b[0] = c[2];
b[1] = c[3];
b[2] = c[2];
b[3] = c[3];
}
clang compiles this with -O3 to:
foo: # @foo
mov rax, qword ptr [rip + c]
mov qword ptr [rip + a], rax
mov rcx, qword ptr [rip + c+8]
mov qword ptr [rip + a+8], rcx
mov qword ptr [rip + a+16], rax
mov qword ptr [rip + a+24], rcx
mov rax, qword ptr [rip + c+16]
mov qword ptr [rip + b], rax
mov rcx, qword ptr [rip + c+24]
mov qword ptr [rip + b+8], rcx
mov qword ptr [rip + b+16], rax
mov qword ptr [rip + b+24], rcx
ret
gcc on the other hand gives the more efficient:
foo:
movapd xmm1, XMMWORD PTR c[rip]
movapd xmm0, XMMWORD PTR c[rip+16]
movaps XMMWORD PTR a[rip], xmm1
movaps XMMWORD PTR a[rip+16], xmm1
movaps XMMWORD PTR b[rip], xmm0
movaps XMMWORD PTR b[rip+16], xmm0
ret
--
You are receiving this mail because:
You are on the CC list for the bug._______________________________________________
llvm-bugs mailing list
[email protected]
http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
