[LLVMbugs] [Bug 12988] New: ARM NEON intrinsic vextq_u16() produces very bad code

[bugzilla-daemon at llvm.org]

http://llvm.org/bugs/show_bug.cgi?id=12988

             Bug #: 12988
           Summary: ARM NEON intrinsic vextq_u16() produces very bad code
           Product: new-bugs
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: simon.hosie at arm.com
                CC: llvmbugs at cs.uiuc.edu
    Classification: Unclassified


Created attachment 8659
  --> http://llvm.org/bugs/attachment.cgi?id=8659
test source

% clang --version
clang version 3.2 (trunk 157465)  
Target: x86_64-unknown-linux-gnu
Thread model: posix

% llc --version
LLVM (http://llvm.org/):
  LLVM version 3.2svn
  Optimized build with assertions.
  Built May 25 2012 (12:07:32).
  Default target: x86_64-unknown-linux-gnu
  Host CPU: penryn
[...]
% clang -ccc-host-triple armv7-none-linux-gnueabi -std=c99 -Werror -O3
-ffast-math -mcpu=cortex-a9 -mfpu=neon -marm -S horriblyvext.c


Rather than emitting a vext instruction, as one might expect, each lane is
stored to memory, one-by-one, and then loaded back into a register.

-- 
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.