Page 3 of 5 FirstFirst 12345 LastLast
Results 21 to 30 of 44

Thread: Ubuntu 11.04: i686 vs. i686 PAE vs. x86_64

  1. #21
    Join Date
    Apr 2010
    Posts
    271

    Default

    Quote Originally Posted by Michael View Post
    For those that don't like the colors, the color algorithm in pts_Graph and bilde_renderer can be improved. Patches welcome. Code at http://www.phoronix-test-suite.com/
    So does this mean you don't see the problem in the coloring? Or just that you don't know enough about the visible spectrum to know what colors to pick?

  2. #22
    Join Date
    Jul 2009
    Posts
    416

    Default

    Quote Originally Posted by fhuberts View Post
    Fedora has had multiarch since the 'beginning'.
    I love the way Fedora and other RPM distros handle multiarch. It's much better than Debian.

    Code:
    yum install $package.i686
    No fiddling around with weird package names like whatever Debian uses for 32-bit libs which isn't guaranteed to exist in the first place.

  3. #23

    Default

    Quote Originally Posted by locovaca View Post
    So does this mean you don't see the problem in the coloring? Or just that you don't know enough about the visible spectrum to know what colors to pick?
    I can see the difference in colors but this color is coming up with an automated color picker off of one base color by rotating it around the color wheel, shading, etc.

    There's something not ideal right now in that code, but it's a low priority item for me to look into especially as I am not too much into colors and have more important work to do, so anyone that is more concerned about colors, patches are welcome.

    phoronix-test-suite/pts-core/objects/bilde_renderer/bilde_renderer.php circa line 450 with color_cache() function is where it can probably best be worked on.

  4. #24

    Default 64 bit kernel/32 bit userland

    I'd be curious how a system with a 64 bit kernel and 32 bit userland does in these comparisons. I've been running a system like this for years and it works great. Seems like the perfect compromise to me.

  5. #25
    Join Date
    Sep 2007
    Location
    Connecticut,USA
    Posts
    962

    Default

    Quote Originally Posted by eikenberry View Post
    I'd be curious how a system with a 64 bit kernel and 32 bit userland does in these comparisons. I've been running a system like this for years and it works great. Seems like the perfect compromise to me.
    Is it simply installing an x86_64 kernel and modules while just keeping the 32 bit userland intact?

  6. #26
    Join Date
    Feb 2008
    Location
    Linuxland
    Posts
    5,128

    Default

    Quote Originally Posted by DeepDayze View Post
    Is it simply installing an x86_64 kernel and modules while just keeping the 32 bit userland intact?
    Yep. It's even a supported mode in Debian, IIRC.

  7. #27
    Join Date
    Oct 2008
    Posts
    3,149

    Default

    Quote Originally Posted by Michael View Post
    I can see the difference in colors but this color is coming up with an automated color picker off of one base color by rotating it around the color wheel, shading, etc.

    There's something not ideal right now in that code, but it's a low priority item for me to look into especially as I am not too much into colors and have more important work to do, so anyone that is more concerned about colors, patches are welcome.

    phoronix-test-suite/pts-core/objects/bilde_renderer/bilde_renderer.php circa line 450 with color_cache() function is where it can probably best be worked on.
    You could pick colors at random and get much better results than the colors picked here.

  8. #28

    Default

    Quote Originally Posted by DeepDayze View Post
    Is it simply installing an x86_64 kernel and modules while just keeping the 32 bit userland intact?
    Yes. That is all there is to it. I use Debian but unless Ubuntu really broke things it should work perfectly fine there as well.

    One area where I've heard people had issues was with the fglrx (ati binary) driver. It's kernel module didn't like the setup for some reason. Not sure if it was a packaging issue or a bug with the driver itself, but something to watch out for.

  9. #29
    Join Date
    Aug 2007
    Posts
    6,634

    Default

    That's an unsupported combination because you need to build the kernel module for 64 bit and the userspace for 32 bit. Usually you need at least a 64 bit chroot to build the 64 bit kernel module in that case. That combination is only needed when you want to install/rescue a 64 bit system via chroot, there is no real good other use case for 64 bit kernel and 32 bit userspace. The things that gain speed are in the 64 bit userspace, the kernel is not that important.

  10. #30

    Default

    Quote Originally Posted by Kano View Post
    That's an unsupported combination because you need to build the kernel module for 64 bit and the userspace for 32 bit. Usually you need at least a 64 bit chroot to build the 64 bit kernel module in that case. That combination is only needed when you want to install/rescue a 64 bit system via chroot, there is no real good other use case for 64 bit kernel and 32 bit userspace. The things that gain speed are in the 64 bit userspace, the kernel is not that important.
    As someone already pointed out, it is a supported setup on Debian. And all the modules build fine on the system as the required 64 bit libraries are installed. So I'm interpreting this as saying that Ubuntu breaks with Debian on this.

    The use case is getting access to more memory w/o the PAE penalty while keeping compatibility with the 32bit binary only userland apps. I thought that would be pretty obvious from the context of this article.

Posting Permissions

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