https://llvm.org/bugs/show_bug.cgi?id=24364
Bug ID: 24364
Summary: [x86, FMA] negated fma pattern not recognized
Product: libraries
Version: trunk
Hardware: PC
OS: All
Status: NEW
Severity: normal
Priority: P
Component: Backend: X86
Assignee: [email protected]
Reporter: [email protected]
CC: [email protected]
Classification: Unclassified
Testing llc built from r244046:
$ cat fnmsub.ll
define float @nmsub(float %a, float %b, float %c) {
%mul = fmul fast float %a, %b
%add = fadd fast float %c, %mul
%neg = fsub fast float 0.0, %add
ret float %neg
}
$ ./llc -o - fnmsub.ll -mattr=fma -enable-unsafe-fp-math
...
vfmadd213ss %xmm2, %xmm1, %xmm0
vxorps LCPI0_0(%rip), %xmm0, %xmm0
retq
-----------------------------------------------------------------
I think this should just be a 'vfnmsub213ss'; we shouldn't have to negate
(vxorps) as a separate operation.
--
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
