Results 1 to 8 of 8

Thread: Xi Graphics' Proprietary X Server, Drivers Have Faded Away

Hybrid View

  1. #1
    Join Date
    Jan 2007
    Posts
    10,361

    Default Xi Graphics' Proprietary X Server, Drivers Have Faded Away

    Phoronix: Xi Graphics' Proprietary X Server, Drivers Have Faded Away

    Xi Graphics, the company that once developed proprietary X Servers and graphics drivers for Linux and UNIX platforms, has faded away...

    http://www.phoronix.com/vr.php?view=MTE1MjQ

  2. #2
    Join Date
    Feb 2008
    Location
    Linuxland
    Posts
    3,572

    Default

    Is there a benchmark for X server overhead? If so, you could bench vesa on both, and the results would show what parts of Xorg need improvement.

  3. #3
    Join Date
    Apr 2010
    Posts
    1,946

    Default

    http://www.xig.com/Pages/Edu/OpenSourceATI.pdf

    They were so in for money, that they did not hestate to criticize opensource projects.
    If this is the criteria of a well-designed X Window Subsystem, then the freeware-
    based Linux X servers from XFree86/X.org in combination with the "drivers" from XFree86/X.org
    and/or the graphics manufacturers fail miserably.
    Based on the following, they made majority of income selling driver for ATI cards.
    The failure of the "open source community" vehicle to provide quality X graphics software suitable
    for use in systems that require stability, good performance, and economical maintainability - often
    over long product life cycles - is not difficult to document. Just download the Linux graphics driver
    software packages from any of the major graphics chip manufacturers and examine them. Or visit
    the Linux user forums at those graphics manufacturers Web sites, although some have been
    discontinued because of the embarrassingly negative comments from users. Also one can delve
    into the Linux v2.6 kernel history. When even RedHat has moved to isolate itself from the
    machinations of the Linux kernel group, one suspects things are a bit less than optimal. Clearly
    the Linux 2.6 kernel has been a "moving target" for quite some time, and the prospects for a
    stable version seem dim.
    Using generic terms, they went on critizing idea about inkernel single driver (DRI) and network-less graphics system Wayland >:/
    The constantly changing Linux kernel coupled with the move by the XFree86/X.org "community"
    over the last few years to move parts of X out of the X server and into the kernel - "Direct
    Rendering Infrastructure" is just one (but big) example - dooms the Open Source "Linux graphics
    drivers" activity to one that has no hope of ever providing stable, high-quality, and easily
    maintained graphics software to the UNIX world. It hasn't happened in the past ten years, and it
    won't happen in the next ten. Apparently some in the open source X community have begun to
    grasp this reality, since there have been calls for X to be "dumped," to be replaced with an entirely
    new approach to graphics on UNIX. After making a hash out of X by ignoring the precepts of
    UNIX, and the specs of X, now they want to junk X. Cute.
    And they also were not looking into developing the Xorg, Kernel etc.
    They made money only by porting their existing drivers (made on the basis of opened ati documentation), to customers who paid - as proprietary solution.
    This would mean, their presence or absence does not help GNU/Linux at all.
    In other words, the X server architecture must be properly designed to support the overall goals of
    particular X applications. Then, the "driver" must efficiently harness the hardware capabilities in
    the specific graphics hardware in a manner that is compatible with the X server.
    And now, they conclude their "epicness" by saying in-kernel graphics solution is insecure, but X-only solution must be ported over and over. And they view this as "ideal"...
    When a graphics manufacturer develops a Linux/UNIX driver for a chip, the driver must be
    designed to "mate up" with the X server that is to be used. Unfortunately, if the X server is to be
    an XFree86 or X.org X server, a number of difficulties are encountered. There are many such X
    servers, and one driver will not work with the majority of them. Once a driver is working with one
    or more of these X servers, the manufacturer finds that the X servers are constantly changing.
    Thus the previously working drivers are often "broken" with a new X server release. Then, the
    Linux Group makes a kernel change, and both the X servers and drivers break. And so on.

    And, concluding, this is the reason why they went bankrupt:
    - Intel started to pay and produce opensource drivers.
    - Nvidia did own closed source drivers, but properly.
    - ATI closed documentation, provided own fglrx driver and later to reopened docs with own opensource solution. **

    They went from market because things improved and their proprietary solution to "un-suck" specific X version was not needed anymore.

    I'd say, its GOOD.


    **
    http://www.xig.com/Pages/CardsChips/...I-GENERAL.html
    Support for the R300 and later chips has been restricted somewhat by a new ATI Management Decision to withhold certain register documentation from Third Parties, information necessary for full support of some cards using these chips.

  4. #4
    Join Date
    May 2012
    Location
    Colombia
    Posts
    14

    Default

    Site updated 26 Oct, 2011
    The site design may be outdated, but the info is not.

  5. #5
    Join Date
    Sep 2007
    Location
    Connecticut,USA
    Posts
    873

    Default

    Do they even have any patents they could use against X.org or Linux in general? If so, they may be rather weak patents that have had prior art.

    I remember reading about Xi Graphics in the late '90s and while they had a decent server then (I played with a trial version), certainly now native Xorg is much better than Xi.

    So yet another proprietary maker bites the dust.

  6. #6
    Join Date
    Aug 2008
    Posts
    68

    Default

    You know the recent security vulnerability in the closed-source NVidia drivers - the one where you could get root by reprogramming the GPU registers in a clever way? From what I could tell from their website at the time, Accelerated-X doesn't even try and protect against this kind of attack - their architecture is based on a kernel module that gives arbitrary user-space applications unrestricted access to the entire GPU. In fact, they even advertise this as a feature!

Posting Permissions

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