[LLVMbugs] [Bug 10580] New: [AVX] vector comparisons are scalarized

[bugzilla-daemon at llvm.org]

http://llvm.org/bugs/show_bug.cgi?id=10580

           Summary: [AVX] vector comparisons are scalarized
           Product: new-bugs
           Version: trunk
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: matt at pharr.org
                CC: llvmbugs at cs.uiuc.edu


Given

define <8 x i1> @foo___ff(<8 x float> %a, <8 x float> %b) nounwind readnone {
  %bincmp = fcmp olt <8 x float> %a, %b
  ret <8 x i1> %bincmp
}

llc -mattr=+avx emits a ton of vector extracts, scalar floating-point compares,
and vector inserts to compute the final result.  (The same is true of integer
compares as well.)

It would be nice for it to emit the single corresponding AVX vector compare
instruction instead.

-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.