[LLVMbugs] [Bug 20647] New: Worse codegen than gcc for heaps of bit arith

[bugzilla-daemon at llvm.org]

http://llvm.org/bugs/show_bug.cgi?id=20647

            Bug ID: 20647
           Summary: Worse codegen than gcc for heaps of bit arith
           Product: libraries
           Version: 3.4
          Hardware: PC
                OS: All
            Status: NEW
          Severity: normal
          Priority: P
         Component: Backend: X86
          Assignee: unassignedbugs at nondot.org
          Reporter: kvanberendonck at gmail.com
                CC: llvmbugs at cs.uiuc.edu
    Classification: Unclassified

Look here for a comparison and related source code side-by-side.

https://gist.github.com/kvanberendonck/3d072ae0f450f20cf1ed

What I've noticed mainly is that while gcc performs the arith directly on the
memory location, llvm produces many more instructions because it chooses to
load it into a register first, perform the arith and then write it back. It
also generates twice as many ORs... not sure why.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20140813/a11dd87f/attachment.html>