Intel Graphics Media Accelerator 950 - Hardware

This is a discussion on Intel Graphics Media Accelerator 950 - Hardware ; Anybody using the subject? Is it easy to install in linux (which distribution?) and does it work fine in 3D games or GL screensaver? -- Jerry Wong http://www.geocities.com/jerrywong.geo 但我*抪茈L的應許、 盼望新天新地,有義居在其中。(彼後3:13) But in keeping with his promise we are looking forward ...

+ Reply to Thread
Results 1 to 14 of 14

Thread: Intel Graphics Media Accelerator 950

  1. Intel Graphics Media Accelerator 950

    Anybody using the subject? Is it easy to install in linux (which
    distribution?) and does it work fine in 3D games or GL screensaver?


    --
    Jerry Wong

    http://www.geocities.com/jerrywong.geo
    但我*抪茈L的應許、 盼望新天新地,有義居在其中。(彼後3:13)
    But in keeping with his promise we are looking forward to
    new heaven and a new earth, the home of righteousness. (2 Peter 3:13)



  2. Re: Intel Graphics Media Accelerator 950

    "Jerry Wong" writes:
    >does it work fine in 3D games or GL screensaver?


    The performance of the GMA 950 is pretty low. Probably good enough
    for Beryl and some older games, but too slow for newer stuff. Even
    planetpenguin-racer will probably be a problem at higher resolutions,
    but might work fine at lower resolutions or lower bpp.

    Glxgears is not a very representative benchmark, but to give you an
    idea, here's what lists for some graphics cards
    with free drivers at 24bpp:

    fps card
    6295 Radeon X850XT
    3503 Radeon 9250
    1500 Intel G965 (GMA3000)
    1467 Intel 945G (GMA950)

    A good reason to buy such a board is to support a hardware company
    that supports free software by providing the necessary hardware
    information and some manpower.

    - anton
    --
    M. Anton Ertl Some things have to be seen to be believed
    anton@mips.complang.tuwien.ac.at Most things have to be believed to be seen
    http://www.complang.tuwien.ac.at/anton/home.html

  3. Re: Intel Graphics Media Accelerator 950

    anton@mips.complang.tuwien.ac.at (Anton Ertl) wrote:
    > fps card
    > 6295 Radeon X850XT
    > 3503 Radeon 9250
    > 1500 Intel G965 (GMA3000)
    > 1467 Intel 945G (GMA950)
    >
    > A good reason to buy such a board is to support a hardware company
    > that supports free software by providing the necessary hardware
    > information and some manpower.


    I only want to point out that "such a board" probably refers to an intel
    board and not the radeon X858XT. Intel has supported the development of
    opensource drivers and ATI did support development of opensource drivers
    for cards up to and including 9250. However, ATI did not support
    development of the opensource drivers for X850XT, the r300 code was
    developed by reverse engineering.

    regards Henrik
    --
    The address in the header is only to prevent spam. My real address is:
    hc1(at)poolhem.se Examples of addresses which go to spammers:
    root@localhost postmaster@localhost


  4. Re: Intel Graphics Media Accelerator 950

    Henrik Carlqvist writes:
    >anton@mips.complang.tuwien.ac.at (Anton Ertl) wrote:
    >> A good reason to buy such a board is to support a hardware company
    >> that supports free software by providing the necessary hardware
    >> information and some manpower.

    >
    >I only want to point out that "such a board" probably refers to an intel
    >board


    Yes. Sorry if that was unclear.

    - anton
    --
    M. Anton Ertl Some things have to be seen to be believed
    anton@mips.complang.tuwien.ac.at Most things have to be believed to be seen
    http://www.complang.tuwien.ac.at/anton/home.html

  5. Re: Intel Graphics Media Accelerator 950

    Dances With Crows wrote:
    >
    > The Intel 9?? cards are a pain in Linux AFAICT.


    At least the GMA 950 (as part of my 945 chipset) works fine. I have a Laptop
    with that chip running very nice on my gentoo linux out of the box (with
    the i810 X driver as well as the slow frame buffer driver). Framebuffer
    console and boot splash are working as well. Indeed, Beryl does NOT really
    work, I get not transparency with it (so it runs, but looks ugly). 3D games
    (e.g. enemy territory) work but slow.


    --
    Sorry for the fake e-mail address, you need to remove ".NO" and "SPAM."
    before sending me any mails...

  6. PCI Vs AGP [was: Intel Graphics Media Accelerator 950]

    Anton Ertl wrote:

    [snip]
    > Glxgears is not a very representative benchmark, but to give you an
    > idea, here's what lists for some graphics cards
    > with free drivers at 24bpp:
    >
    > fps card
    > 6295 Radeon X850XT
    > 3503 Radeon 9250

    [snip]

    My Radeon 9250 is only giving 460 FPS. Is that because it is a
    PCI card, rather than an AGP? Or have I just hopelessly
    miss-configured it?


    --
    sig goes here...
    Peter D.

  7. Re: PCI Vs AGP [was: Intel Graphics Media Accelerator 950]

    "Peter D." writes:
    >Anton Ertl wrote:
    >
    >[snip]
    >> Glxgears is not a very representative benchmark, but to give you an
    >> idea, here's what lists for some graphics cards
    >> with free drivers at 24bpp:
    >>
    >> fps card
    >> 6295 Radeon X850XT
    >> 3503 Radeon 9250

    >[snip]
    >
    >My Radeon 9250 is only giving 460 FPS. Is that because it is a
    >PCI card, rather than an AGP?


    I doubt it.

    >Or have I just hopelessly
    >miss-configured it?


    My guess would be that you don't have hardware acceleration for some
    reason, probably because the software version you run does not include
    Mesa DRI R200. You can see that by running the script from
    http://free3d.org/, in particular:

    glxinfo | egrep -A2 "direct rendering|OpenGL vendor"

    Another reason could be that you are running glxgears at more than the
    default size.

    - anton
    --
    M. Anton Ertl Some things have to be seen to be believed
    anton@mips.complang.tuwien.ac.at Most things have to be believed to be seen
    http://www.complang.tuwien.ac.at/anton/home.html

  8. Re: PCI Vs AGP [was: Intel Graphics Media Accelerator 950]

    Anton Ertl wrote:

    > "Peter D." writes:
    >>Anton Ertl wrote:
    >>
    >>[snip]
    >>> Glxgears is not a very representative benchmark, but to give you an
    >>> idea, here's what lists for some graphics cards
    >>> with free drivers at 24bpp:
    >>>
    >>> fps card
    >>> 6295 Radeon X850XT
    >>> 3503 Radeon 9250

    >>[snip]
    >>
    >>My Radeon 9250 is only giving 460 FPS. Is that because it is a
    >>PCI card, rather than an AGP?

    >
    > I doubt it.
    >
    >>Or have I just hopelessly
    >>miss-configured it?

    >
    > My guess would be that you don't have hardware acceleration for some
    > reason, probably because the software version you run does not include
    > Mesa DRI R200. You can see that by running the script from
    > http://free3d.org/, in particular:
    >
    > glxinfo | egrep -A2 "direct rendering|OpenGL vendor"


    ->glxinfo | egrep -A2 "direct rendering|OpenGL vendor"
    libGL warning: 3D driver claims to not support visual 0x4c
    direct rendering: Yes
    server glx vendor string: SGI
    server glx version string: 1.2
    --
    OpenGL vendor string: Tungsten Graphics, Inc.
    OpenGL renderer string: Mesa DRI R200 20060327 x86/MMX+/3DNow!+/SSE2 TCL
    OpenGL version string: 1.3 Mesa 6.5

    Mesa DRI R200 is in there.

    > Another reason could be that you are running glxgears at more than the
    > default size.


    460 FPS without me deliberately changing the size, I guess that
    it is possible that there is some size configuration going on
    that I can not see.

    720 FPS when I shrink it as far as I can.

    ~4000 FPS when the glxgears window is covered up.


    --
    sig goes here...
    Peter D.

  9. Re: PCI Vs AGP [was: Intel Graphics Media Accelerator 950]

    "Peter D." writes:
    >Anton Ertl wrote:
    >
    >> "Peter D." writes:
    >>>My Radeon 9250 is only giving 460 FPS. Is that because it is a
    >>>PCI card, rather than an AGP?

    ....
    >OpenGL renderer string: Mesa DRI R200 20060327 x86/MMX+/3DNow!+/SSE2 TCL
    >OpenGL version string: 1.3 Mesa 6.5
    >
    >Mesa DRI R200 is in there.
    >
    >> Another reason could be that you are running glxgears at more than the
    >> default size.

    >
    >460 FPS without me deliberately changing the size, I guess that
    >it is possible that there is some size configuration going on
    >that I can not see.


    Not with the glxgears versions that I have seen. They all default
    to some small (128x128?) window.

    Maybe it is really the difference between AGP and PCI; the bandwidth
    of PCI should be sufficient for glxgears, but maybe the absence of
    some AGP feature (e.g., the GART) disables some speed-critical part of
    the driver. I remember that the GART size had a significant effect on
    UT2004 (don't remember if there was an effect on glxgears).

    - anton
    --
    M. Anton Ertl Some things have to be seen to be believed
    anton@mips.complang.tuwien.ac.at Most things have to be believed to be seen
    http://www.complang.tuwien.ac.at/anton/home.html

  10. Re: PCI Vs AGP [was: Intel Graphics Media Accelerator 950]

    Peter D. wrote:
    > direct rendering: Yes


    Here is my output on an AGP Radeon 9000, using the open source driver:

    mark@venus:~$ glxgears
    8203 frames in 5.0 seconds = 1640.600 FPS
    8485 frames in 5.0 seconds = 1697.000 FPS
    11782 frames in 5.0 seconds = 2356.400 FPS
    11783 frames in 5.0 seconds = 2356.600 FPS
    11784 frames in 5.0 seconds = 2356.800 FPS

    There does appear to be something wrong with your setup. What distro are you
    using and what kernel version?

    --
    Mark Hobley
    393 Quinton Road West
    QUINTON
    Birmingham
    B32 1QE

    Telephone: (0121) 247 1596
    International: 0044 121 247 1596

    Email: markhobley at hotpop dot donottypethisbit com

    http://markhobley.yi.org/


  11. Re: PCI Vs AGP [was: Intel Graphics Media Accelerator 950]

    Mark Hobley wrote:

    > Peter D. wrote:
    >> direct rendering: Yes

    >
    > Here is my output on an AGP Radeon 9000, using the open source driver:
    >
    > mark@venus:~$ glxgears
    > 8203 frames in 5.0 seconds = 1640.600 FPS
    > 8485 frames in 5.0 seconds = 1697.000 FPS
    > 11782 frames in 5.0 seconds = 2356.400 FPS
    > 11783 frames in 5.0 seconds = 2356.600 FPS
    > 11784 frames in 5.0 seconds = 2356.800 FPS
    >
    > There does appear to be something wrong with your setup. What distro are
    > you using and what kernel version?


    Mandriva 2007.0

    I was using a custom kernel, but I've just rebooted with a Mandriva
    kernel; 2.6.17-10mdv, and I've rerun glxgears.

    ->glxgears
    libGL warning: 3D driver claims to not support visual 0x4c
    2353 frames in 5.0 seconds = 470.551 FPS
    2344 frames in 5.0 seconds = 468.777 FPS
    2340 frames in 5.0 seconds = 467.809 FPS



    --
    sig goes here...
    Peter D.

  12. Re: PCI Vs AGP [was: Intel Graphics Media Accelerator 950]

    Anton Ertl wrote:

    > "Peter D." writes:
    >>Anton Ertl wrote:
    >>
    >>> "Peter D." writes:
    >>>>My Radeon 9250 is only giving 460 FPS. Is that because it is a
    >>>>PCI card, rather than an AGP?

    > ...
    >>OpenGL renderer string: Mesa DRI R200 20060327 x86/MMX+/3DNow!+/SSE2 TCL
    >>OpenGL version string: 1.3 Mesa 6.5
    >>
    >>Mesa DRI R200 is in there.
    >>
    >>> Another reason could be that you are running glxgears at more than the
    >>> default size.

    >>
    >>460 FPS without me deliberately changing the size, I guess that
    >>it is possible that there is some size configuration going on
    >>that I can not see.

    >
    > Not with the glxgears versions that I have seen. They all default
    > to some small (128x128?) window.


    That is about what it looks like.

    > Maybe it is really the difference between AGP and PCI; the bandwidth
    > of PCI should be sufficient for glxgears, but maybe the absence of
    > some AGP feature (e.g., the GART) disables some speed-critical part of
    > the driver. I remember that the GART size had a significant effect on
    > UT2004 (don't remember if there was an effect on glxgears).


    ->lsmod | grep -i agp
    amd64_agp 10948 0
    agpgart 27240 2 drm,amd64_agp


    --
    sig goes here...
    Peter D.

  13. Re: PCI Vs AGP [was: Intel Graphics Media Accelerator 950]

    "Peter D." writes:
    >Anton Ertl wrote:
    >> Maybe it is really the difference between AGP and PCI; the bandwidth
    >> of PCI should be sufficient for glxgears, but maybe the absence of
    >> some AGP feature (e.g., the GART) disables some speed-critical part of
    >> the driver.


    Yes, that seems to be it. I just forced my Radeon 9600 AGP into PCI
    mode with

    Option "BusType" "PCI"

    and the glxgears output dropped from 2300fps to 800fps (that the
    factor is worse for the 9250 may be due to the difference between the
    9600 and the 9250 (or the r300 vs. r200 drivers), or due to the fact
    that there is something else going on in a real PCI card; I'm not
    going to swap graphics cards to find that out).

    Note that you cannot put a PCI card in AGP mode.

    BTW, concerning the GART, even with the option above my
    /var/log/Xorg.0.log still mentions the GART several times, so
    apparently the PCI mode is not perfect.

    >->lsmod | grep -i agp
    >amd64_agp 10948 0
    >agpgart 27240 2 drm,amd64_agp


    Yes, the memory controller has GART support, but with the PCI card
    this is not used. I am not an expert on this, but AGP cards support
    at least one additional way of transferring data between main memory
    and the graphics card that is not available on PCI cards (and it has
    to do with the GART).

    If the driver makes use of that (and it appears it does) and the
    general (i.e., including PCI) way of doing things is not very
    efficient (either because it was not optimized as much, or because
    it's not possible (I doubt that the latter is the case for glxgears)),
    then PCI performance suffers.

    - anton
    --
    M. Anton Ertl Some things have to be seen to be believed
    anton@mips.complang.tuwien.ac.at Most things have to be believed to be seen
    http://www.complang.tuwien.ac.at/anton/home.html

  14. Re: PCI Vs AGP [was: Intel Graphics Media Accelerator 950]

    Anton Ertl wrote:

    > "Peter D." writes:
    >>Anton Ertl wrote:
    >>> Maybe it is really the difference between AGP and PCI; the bandwidth
    >>> of PCI should be sufficient for glxgears, but maybe the absence of
    >>> some AGP feature (e.g., the GART) disables some speed-critical part of
    >>> the driver.

    >
    > Yes, that seems to be it. I just forced my Radeon 9600 AGP into PCI
    > mode with
    >
    > Option "BusType" "PCI"
    >
    > and the glxgears output dropped from 2300fps to 800fps

    [snip]

    Thanks Anton.


    --
    sig goes here...
    Peter D.

+ Reply to Thread