Announcement

Collapse
No announcement yet.

AMD Catalyst 11.11 Brings Critical Linux Changes

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • #61
    "This means the recently-released Fedora 11 should now work."

    Too many elevens in article.

    (Also, the link is not working for me; says "file not found")


    Comment


    • #62
      Originally posted by Massa View Post
      - can't suspend on Arch Linux. I have Debian sid on a different partition and that one suspends fine though.
      This was my fault, I forgot to properly disable KMS/radeon. Fixed now!

      Comment


      • #63
        It seems, that the tearing has gotten more worse than in 11.10.

        Will AMD every fix this problem? And again there is something big announced for the 11.12 catalyst. Thank you so much! STOP implementing new features, and start finishing the simpliest driverfeatures! the old driver isn't even usable in 2D environment... That sucks so hard!

        BTW using a HD5770. Nvidia has to announce the new gtx-generation. Then AMD will never see any more cash from my side!

        Comment


        • #64
          Originally posted by psuch View Post
          Will AMD every fix this problem? And again there is something big announced for the 11.12 catalyst. Thank you so much! STOP implementing new features, and start finishing the simpliest driverfeatures! the old driver isn't even usable in 2D environment... That sucks so hard!

          BTW using a HD5770
          haha, u sound like me when I got my hd5770 which I guess was about two years ago now.....and, while things are better in catalyst now than then, its not by very much.

          They implemented the tear free option earlier this year, which does seem to solve most tearing, however it takes up what seems to be most of ur cards onboard memory which causes issue when trying to do gpu intensive things, and worse still I have always experienced some degree of a micro-lag. This lag shows itself whenever I play HD videos, and since the dawning of unity on ubuntu (which I use only because I hate gnome 3, and am too lazy to figure out another desktop for the time being) all compiz effects, which are already much slower than releases past, are often choppy. Oh yeah, and with tear free on scrolling in a maximized browser on my 1080p display is just horrendously slow and choppy.

          So if u ask me, what u really want to happen, at least as long as u have this card, is for dynamic power management in the open source driver to be rewritten so that it actually works, because the open source drivers, while miles behind in its 3d features (but catching up), is absolutely a superior experience. If they didn't make my cards fan spin like a bastard, and force me to manually choose everytime I need to change the power setting I would definately be using it right now

          Comment


          • #65
            Originally posted by psuch View Post
            It seems, that the tearing has gotten more worse than in 11.10.

            Will AMD every fix this problem? And again there is something big announced for the 11.12 catalyst. Thank you so much! STOP implementing new features, and start finishing the simpliest driverfeatures! the old driver isn't even usable in 2D environment... That sucks so hard!

            BTW using a HD5770. Nvidia has to announce the new gtx-generation. Then AMD will never see any more cash from my side!
            Would be nice to have more details about the system you're running and what you've done to come to this conclusion. Also information about why you've ruled out a botched installation.

            Comment


            • #66
              Xvideo / xv output bug

              Originally posted by paproch View Post
              @Kano


              This is xv issue.
              This seems to be confirmed by Debian bug #649346. I guess this is the natural consequence of fglrx being too behind the development curve from Xorg? Apparently this xv bug is not a problem on downgraded Xorg (1.10, I think).

              I would run the open source radeon driver but my machine, HP 8460w, does not allow changes to backlight brightness with open source driver (via sysfs). Also it is my understanding that VA-API is a no-go on open source radeon driver. How silly that we cannot have working features in desktop Linux.

              Does anyone know if it is possible to use Sandy Bridge VA acceleration with a discrete graphics system? I'm guessing this is impossible because of memory address writing issues across different chips / framebuffer spaces. For now I only have accelerated video decode on GL output module (I use VLC), or on X11 video output, and these consume 100% and ~50% CPU respectively. How stupid.

              Comment


              • #67
                Xvideo / xv output bug

                Originally posted by paproch View Post
                @Kano


                This is xv issue.
                This seems to be confirmed by Debian bug #649346. I guess this is the natural consequence of fglrx being too behind the development curve from Xorg? Apparently this xv bug is not a problem on downgraded Xorg (1.10, I think).

                I would run the open source radeon driver but my machine, HP 8460w, does not allow changes to backlight brightness with open source driver (via sysfs). Also it is my understanding that VA-API is a no-go on open source radeon driver. How silly that we cannot have working features in desktop Linux.

                Does anyone know if it is possible to use Sandy Bridge VA acceleration with a discrete graphics system? I'm guessing this is impossible because of memory address writing issues across different chips / framebuffer spaces. For now I only have accelerated video decode on GL output module (I use VLC), or on X11 video output, and these consume 100% and ~50% CPU respectively. How stupid.

                Comment


                • #68
                  Originally posted by jaidormi View Post
                  This seems to be confirmed by Debian bug #649346. I guess this is the natural consequence of fglrx being too behind the development curve from Xorg? Apparently this xv bug is not a problem on downgraded Xorg (1.10, I think).

                  I would run the open source radeon driver but my machine, HP 8460w, does not allow changes to backlight brightness with open source driver (via sysfs). Also it is my understanding that VA-API is a no-go on open source radeon driver. How silly that we cannot have working features in desktop Linux.

                  Does anyone know if it is possible to use Sandy Bridge VA acceleration with a discrete graphics system? I'm guessing this is impossible because of memory address writing issues across different chips / framebuffer spaces. For now I only have accelerated video decode on GL output module (I use VLC), or on X11 video output, and these consume 100% and ~50% CPU respectively. How stupid.
                  As far as your backlight goes, it may be that it's like mine. It requires some of the ACPI stuff turned on and is all done through that interface rather than through the graphics driver if you're using the open source drivers. As far as VA-API goes, there's no real working support. I've gotten the VDPAU stuff working for mpeg1/2 but it's still sub par (some decoding/rendering errors from some files), this is mostly because the UVD2 stuff has no documentation at all from AMD/ATI in case it could cause someone to be able to bypass DRM restrictions on other operating systems. The situation sucks, but I've found the open source drivers to be far more stable for me than the fglrx ones, even if i don't get as good of 3d performance (and now that fglrx has started to work on 2d, comparable performance).

                  Comment


                  • #69
                    Originally posted by simcop2387 View Post
                    As far as your backlight goes, it may be that it's like mine. It requires some of the ACPI stuff turned on and is all done through that interface rather than through the graphics driver if you're using the open source drivers. As far as VA-API goes, there's no real working support. I've gotten the VDPAU stuff working for mpeg1/2 but it's still sub par (some decoding/rendering errors from some files), this is mostly because the UVD2 stuff has no documentation at all from AMD/ATI in case it could cause someone to be able to bypass DRM restrictions on other operating systems. The situation sucks, but I've found the open source drivers to be far more stable for me than the fglrx ones, even if i don't get as good of 3d performance (and now that fglrx has started to work on 2d, comparable performance).
                    I'm not sure what you mean about the backlight. For me I am attempting to change backlight through /sys/class/backlight regardless of the driver. With fglrx it works, with radeon it doesn't.

                    What distro and versions are you using in your VA API attempts? It basically works for me on the latest Debian packages with 'vlc --ffmpeg-hw,' save for the xv issue folks are experiencing.

                    I've also found radeon > fglrx, but there's not much I can do if I can't get simple backlight control working with the FOSS driver. I'm neither knowledgeable nor skilled enough to reverse engineer what fglrx is doing to enable backlight. Hopefully someone will end up owning this specific laptop who is also a good system level program. But it ain't me babe, it ain't me.

                    Cheers.

                    Comment

                    Working...
                    X