3 Replies Latest reply on Mar 6, 2015 8:42 AM by parvez.subhedar Branched to a new discussion.

    Bug Report for driver 14.9

    joshklint

      On Windows some users experience a "pure virtual function call" error on startup:

      http://www.leadwerks.com/werkspace/topic/11037-leadwerks-33-beta-is-crashing-on-start/

       

      According to this user, 14.9 and 14.9.1 beta drivers work correctly, and beta drivers 14.9.2 cause a crash.

       

      On Linux, the 14.9 driver causes our application to crash under some conditions.  All persons experiencing this problem have AMD GPUs:

      http://www.leadwerks.com/werkspace/topic/11038-linux-project-manager-causes-ide-crash/

       

      I am unable to produce any of these errors with an R9 280 on Windows 7 or Ubuntu 12.04.  I can provide some Steam keys for the application if the driver team doesn't mind using a Steam account to test it.  The application will install for both Windows and Linux.

       

      The problems on Linux appear bad enough I have temporarily removed support for AMD graphics on Linux from our Steam page.

       

      Our application makes use of most OpenGL 4 features, so it might be a helpful tool for future testing before releases.

        • Re: Bug Report for driver 14.9
          pchiruma

          Fix will be given in the next catalyst build

            • Re: Bug Report for driver 14.9
              rexhunter99

              Any news on this front? The bug seems to be specifically related to buffer usage in a multithreaded context. I was using a secondary thread to create buffers in memory and then upload them to the graphics memory, however had to go back to forcing all my GL code into the main thread so my app would work on AMD hardware.

                • Re: Bug Report for driver 14.9

                  0: kd> !analyze -v

                   

                  *******************************************************************************

                   

                  *

                  *

                   

                  •                        Bugcheck

                  Analysis                                    *

                   

                  *

                                               *

                   

                  *******************************************************************************

                   

                   

                   

                  SYSTEM_THREAD_EXCEPTION_NOT_HANDLED (7e)

                   

                  This is a very common bugcheck.  Usually the exception address pinpoints

                   

                  the driver/function that caused the problem.  Always note this address

                   

                  as well as the link date of the driver/image that contains this address.

                   

                  Arguments:

                   

                  Arg1: ffffffffc0000005, The exception code that was not handled

                   

                  Arg2: fffff800a732708b, The address that the exception occurred at

                   

                  Arg3: ffffd000f1214128, Exception Record Address

                   

                  Arg4: ffffd000f1213930, Context Record Address

                   

                   

                   

                  Debugging Details: