Results 1 to 7 of 7

Thread: GCC 4.6 Brings Intel Optimizations, AVX, Improvements

Hybrid View

  1. #1
    Join Date
    Jan 2007
    Posts
    15,652

    Default GCC 4.6 Brings Intel Optimizations, AVX, Improvements

    Phoronix: GCC 4.6 Brings Intel Optimizations, AVX, Improvements

    While no release announcement has yet to hit the wire, GCC 4.6.0 is now available. Uploaded to the GCC mirrors yesterday afternoon was the GCC 4.6.0 final source packages, right ahead of their planned release which is/was expected to be around Monday. It was only earlier this month that the first release candidate arrives, but this is one rather nice update to the GNU Compiler Collection...

    http://www.phoronix.com/vr.php?view=OTI1NQ

  2. #2
    Join Date
    Jul 2008
    Location
    Greece
    Posts
    3,801

    Default

    Quote Originally Posted by phoronix View Post
    There's also (finally) Intel Core 2 optimizations for GCC that can be tapped using core2 with the mtune and march options.
    Could you give a short explanation here about what you mean? -march=core2 has been working for a long time now.

  3. #3
    Join Date
    Jul 2008
    Location
    Greece
    Posts
    3,801

    Default

    Also, it has already been pointed out to you in another thread that it's "-Ofast", not "-0fast".

  4. #4

    Default

    No matter how I compile unrar with GCC 4.6 the result ends up slower than what GCC 4.5.2 generates.

    Not only slower, the resulting binary is larger (just by 4KB, or about 2%).

  5. #5
    Join Date
    Oct 2009
    Posts
    845

    Default

    Quote Originally Posted by RealNC View Post
    Could you give a short explanation here about what you mean? -march=core2 has been working for a long time now.
    Yes I reacted at that aswell, maybe he meant that there has been additional optimizations added to core2 by CodeSourcery/Intel which have made it into 4.6. But like you said 'core2' has been available through '-march' and '-mtune' for gcc 4.5.x. In fact when I use '-march=native' on my core i5 machine, gcc 4.5 chooses 'core2' as evident by doing a 'gcc -c -Q -march=native --help=target'.

  6. #6
    Join Date
    Dec 2007
    Location
    Edinburgh, Scotland
    Posts
    597

    Default

    There's a good change you sole GNU's thunder by announcing the release before the project did

  7. #7
    Join Date
    Jan 2008
    Location
    Radoboj, Croatia
    Posts
    155

    Default

    Quote Originally Posted by Michael
    I've been running some GCC 4.6 benchmarks versus GCC 4.5.2 along with LLVM 2.9 SVN and the LLVM Clang compiler and LLVM's DragonEgg optimizer for GCC. Those benchmarks should be on Phoronix this coming Monday.
    Could you please add ICC benchmark to your results? I've been using ICC recently and I have a feeling that it generates a lot faster code than GCC.
    I'm using GCC 4.5.2 and ICC 12.0.2
    Code:
    $ gcc --version
    gcc (GCC) 4.5.2 20110127 (prerelease)
    Copyright (C) 2010 Free Software Foundation, Inc.
    This is free software; see the source for copying conditions.  There is NO
    warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
    
    $ icc --version
    icc (ICC) 12.0.2 20110112
    Copyright (C) 1985-2011 Intel Corporation.  All rights reserved.

Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •