[LLVMbugs] [Bug 14867] New: ARM poor vector code generation zext 8xi8 to 8xi32

bugzilla-daemon at llvm.org bugzilla-daemon at llvm.org
Tue Jan 8 22:13:17 PST 2013


             Bug #: 14867
           Summary: ARM poor vector code generation  zext 8xi8 to 8xi32
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: Backend: ARM
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nrotem at apple.com
                CC: llvmbugs at cs.uiuc.edu
    Classification: Unclassified

The LL code below is compiled to a really long sequence of ARM instructions
(thumb/swift). There must be a better way to lower this code. 

%T0_11 = type <8 x i8>                                                    
%T1_11 = type <8 x i32>
define void @func11() {
   %v0 = load %T0_11* undef                                                
   %r = sext %T0_11 %v0 to %T1_11                                          
   store %T1_11 %r, %T1_11* undef
   ret void

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.

More information about the llvm-bugs mailing list