mike lewis wrote: > On 1/13/07, Duncan Webb <[EMAIL PROTECTED]> wrote: >> mike lewis wrote: >>> On 1/13/07, mike lewis <[EMAIL PROTECTED]> wrote: >>>> On 1/13/07, Duncan Webb <[EMAIL PROTECTED]> wrote: >>>>> Lucian Muresan wrote: >>>>>> Jason Tackaberry wrote: >>>>>>> On Fri, 2007-01-12 at 23:31 +0800, mike lewis wrote: >>>>>>>> Before I go off to bed I just have to say that I figured out what the >>>>>>>> issue was. vidix wasn't detecting the RAm size of my card. So I hard >>>>>>>> coded the RAM size (from 16 to 32) and now there is not more artifacts >>>>>>>> on the screen. >>>>>>> Where did you hardcode this? >>>>>> I think the proper way is not to hardcode it, but to apply the so-called >>>>>> "matroxfb-full-memory" kernel patch. Another useful patch might be the >>>>>> one called "matroxfb-g400-clock", I don't know why they haven't made it >>>>>> into the kernel for such a long time. You can find older versions of >>>>>> them in the 'patches' directory of the DirectFB source, and I guess they >>>>>> apply to recent kernels too, or you can easily adapt them by hand. If >>>>>> you prefer, you can use mine attached in this message, already adapted >>>>>> for 2.6.18 (should also apply to 2.6.19), the one called >>>>>> linux-2.6.18_matroxfb-address-ioctl_v3.diff is necessary for newer >>>>>> versions of DirectFB. >>>>> I wouldn't use the "matroxfb-g400-clock" patch, Villi recommended to use >>>>> it and I tried it recently and it fucked up the driver big time. May be >>>>> it depends on the card a bit, mines a DH with sdram. >>>>> >>>>> The "matroxfb-full-memory" is good though. >>>>> >>>> Hmm, Not sure about this one.. I'm not useing FB just mga_vid under X?? >>>> >>> I just had another look at mga_vid.c in under vidix and I still can't >>> work out how they are working out the RAM size. But I'm sure these >>> two feature (matroxfb and mga_vid) are not one in the same... >> All depends if the mga_vid driver sits on top of the framebuffer or >> access the hardware directly. If it sits on the framebuffer then the >> full memory patch should make a difference. >> >> Having said this a quick look at the code and it seems that mga_vid >> access the hardware directly. From my configuration I see: >> # dmesg | grep mga >> mga_vid: Found MGA G400/G450 at 0000:01:00.0 >> mga_vid: MMIO at 0xe39e0000 framebuffer: 0xF6000000 >> mga_vid: OPTION word: 0x40040120 mem: 0x00 SDRAM >> mga_vid: detected RAMSIZE is 16 MB >> mga_vid: 1 supported cards found >> mga_vid: using major: 83 (assigned or default!) >> >> So it would seem that my card has 16MB of SDRAM. which is a bit strange >> as the memory chips are SGRAM. This has determined by the PCI config of >> the card. May be updating the card's firmware changed this. >> >> The size can be overridden by the modprobe options: >> # modinfo mga_vid >> filename: /lib/modules/2.6.18.5-hawk-r1/extra/mga_vid.ko >> author: Aaron Holtzman <[EMAIL PROTECTED]> >> license: GPL >> parmtype: mga_ram_size:array of int >> parmtype: mga_top_reserved:array of int >> parmtype: mga_brightness:array of int >> parmtype: mga_contrast:array of int >> parmtype: major:int >> vermagic: 2.6.18.5-hawk-r1 preempt mod_unload PENTIUMIII gcc-3.4 >> depends: >> >> I think that a: >> modprobe mga_vid mga_ram_size=32 >> should do the trick then I see: >> mga_vid: Found MGA G400/G450 at 0000:01:00.0 >> mga_vid: MMIO at 0xe39e0000 framebuffer: 0xF6000000 >> mga_vid: OPTION word: 0x40040120 mem: 0x00 SDRAM >> mga_vid: RAMSIZE forced to 32 MB >> mga_vid: 1 supported cards found >> mga_vid: using major: 83 (assigned or default!) >> >> > Yep, not problems there. Mine is a problem with xine's implementation > of the mga_vid vo in vidix. > > --- > [EMAIL PROTECTED]:~$ dmesg | grep mga > [17179594.884000] mga_vid: Found MGA G400/G450 at 0000:01:00.0 > [17179594.884000] mga_vid: MMIO at 0xd0c18000 framebuffer: 0xCC000000 > [17179594.884000] mga_vid: OPTION word: 0x50044120 mem: 0x10 SGRAM > [17179594.884000] mga_vid: detected RAMSIZE is 32 MB > [17179594.884000] mga_vid: 1 supported cards found > [17179594.884000] mga_vid: using major: 83 (assigned or default!) > [17179627.332000] [drm] Initialized mga 3.2.1 20051102 on minor 0 > --- > > but.. > --- > sudo xine -V vidix /mnt/hdb1/Movies/kids/Movie.avi --verbose=2 | grep mga > [mga_crtc2] Found MGA G400/G450 > [mga_crtc2] detected RAMSIZE is 16 MB > [mga_crtc2] Set write-combining type of video memory > [mga_crtc2] IRQ support disabled
IIRC the G400 has only 16MB for the crtc2 (second head) so this isn't a problem. 16MB is plenty for triple buffered tv out at 32bits. Duncan ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ Freevo-devel mailing list Freevo-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freevo-devel
