1 Reply Latest reply on Aug 18, 2011 8:23 AM by abittner

    gdi quality of winxp driver (catalyst display driver alone) is very poor to nonexistant

    abittner
      dont know where to turn elsewhere, as amd bugreports and gameforum reports stay unrespected

      hello amd developer folks and forum fellow participants,

       

      to make a long story very short, the gdi quality of the winxp (32bit) catalyst drivers ever since the 10.6 level (june 2010 was fine, everything else beyond that gives BSOD and crash/freezes) is a catastrophy.

       

      there are numerous threads on the gameforums part, and a very easy benchmark/test that i found out to demonstrate these blue screens of death (BSOD) and freezes. everything running and testing at my side on rather current integrated chipsets 4250, 4270 and so on, all with 32bit winxp professional and drivers including catalyst 10.6 and every release since then, that gives us now  a full year of faulty drivers on the windowsxp 32bit platform.

       

      i have tried to file official bugreports with amd/ati folks, there is a lot of frustration and ranting over in various atigameforum threads and areas of many people there writing about the same and very similar problems, the drivers on winxp even create crash/stackreports that are being saved and need to be sent to some ati.com mailadress that doesnt even accept emails (any more? ever?). i have tried everything to contact official amd ati personel before, to no avail. this is all very frustrating.

       

      microsoft officially supports and sticks to windowsxp 32bit edition at least til middle of 2014, thats still a long way, and these problems of catastrophic driver experience of amd ati products in the gpu business is unfortunately not new. with every new driver release now and then bad things had been introduced, but this gdi problems of simple drawing rectangles, boxes, filled, unfilled, splines and things that exist as gdi commands at least since the ages of windows95 (nintey five, thats right) is really ridiculous.

       

      stuff that has been in the windows world for over 16years now is severly broken.

       

      so i am trying to find my luck here and get some people to see the details and check if this is really the best that amd ati is able to do. maybe someone will sort out these severe gdi glitches of the win32 driver stack after all, i have not yet given up hope, but i am close to do so :(

       

      please do take a thorough look at gameforum thread:

      http://forums.amd.com/game/messageview.cfm?catid=279&threadid=143883

       

      and find my postings and finding there. sorry about all the rants there at least of my part, but it was really a pest and is simply unbearable still until todays date.

       

      hopefully someone does take this message and the catalyst gamereports and technical bugreports seriously.

       

      thanks in advance, and regards.

      p.s. rough summary: the benchmark used for easy and uncomplicated testing is the tomshardware 2dbench.exe, also given as download locations at tomshardware over there in that thread in my posts and elsewhere. the crashes and freezes that occur ever since 10.6+ (so starting with 10.7 and beyond including until 11.7 driver level) are in the third test of simply gdi functions, where the 2dbench.exe wants to draw simple filled rectangles onto the screen. the screen starts bleeding boxes outside the actual drawing area with all the drivers since 10.7 in different slight variations, and then eventually the whole machine freezing more or less quickly, some driver levels making it reboot automatically, some just stay with a black screen after a while and so on.

       

      all of this does NOT happen with the driver version 10.6, the whole 2dbench.exe works flawlessly and repeatedly on this diver level. so once on a faulty driver level, and going back to version 10.6 again, cures the symptoms and the problem.

       

      thanks for listening.

        • gdi quality of winxp driver (catalyst display driver alone) is very poor to nonexistant
          abittner

          still the very same bug behavior with winxp 11.8 driver stack.

          btw, i have overlooked that tomshardware 2dbench can be started more granularily to do separate tests, the textdrawing, the lines, rectangles, circles and so on.

           

          also with 11.8, the text-test (first test), and the last three tests (arc/ellipse, blitting, stretching) all work normally. but all other tests, the second to the fifth all instant-crash, hang or reboot the whole machine. these tests are line, splines/bezier, polygon and rectangle.

           

          on the tomshardware pages to this 2dbench, there is even the exact gdi api calls given that these tests all use, so you amd ati folks can figure out what exactly is wrong in the implementation of those api calls in your driver components ever since 10.7 included and upwards til todays 11.8 driver

           

          tomshardware:

          http://www.tomshardware.com/reviews/2d-windows-gdi,2547-5.html

          quote from the lin drawing benchmark webpage:GDI Functions Called

          For the direct line drawing benchmark:

          CreatePen
          MoveToEx
          LineTo
          SelectObject
          DeleteObject

          For drawing lines from the DIB buffer:

          CreateCompatibleDC
          CreateDIBSection
          CreateSolidBrush
          Rectangle
          BitBlt

          --------

           

          and so on....

          http://www.tomshardware.com/reviews/2d-windows-gdi,2547-7.html

           

          please take a look into these severe gdi bugs in your winxp 32bit drivers. thank you.