No announcement yet.

3D test fails to run after 127 iterations

  • Filter
  • Time
  • Show
Clear All
new posts

  • 3D test fails to run after 127 iterations

    I am attempting to run Performance Test for an extended period of time (12-24hrs) using a script to loop for a few hundred iterations. On the platform that I am testing, this works except that once the 3D test runs for the 127th time, it fails to start again. All the other tests in the script (CPU all, 2D all, mem, disk) continue to operate for the remainder of the iterations.

    The system is running an Ivy Bridge i7 3555LE with the integrated graphics HD 4000. After seeing this issue, I was able to replicate it using a different platform that with a Sandy Bridge and integrated HD 3000 graphics. Both platforms are running Windows 7 pro x64, with Performance test 7.0(1025) 64bit, and 3D always stops right around 127 iterations.

    My first guess is that there is a memory leak somewhere that is preventing the 3D test to continue to operate. Does anyone have any thoughts?


  • #2
    Are you doing this with a DOS script?
    So does PerformanceTest exit & restart between each iteration?

    Yes, I would think a memory leak or resource leak is possible. Might be in the video card driver.

    The Intel drivers are known to have memory leaks when doing 3D.


    • #3
      Thanks for the reply David. I am running the script from inside Performance test using the "execute script" operation, so Performance Test does not exit and restart. I will try setting up a script that exits and restarts the application.


      • #4
        I ran a batch file that opens and closes PerformanceTest for each iteration performed. The 3D never crashes after hundreds of executions. Closing PerformanceTest must clear the problem. Is Passmark in contact with Intel over these memory leaks?


        • #5
          We have made no attempt to tell Intel about memory leaks. The problem is that it isn't easy to know with certainty if your particular leak is in the driver, in DirectX, in the operating system, or in our code. It would take some day to track down. Considering we have had only 1 report of this problem, and PerformanceTest V8 will be released in the next few days it doesn't make sense to spend days looking at the problem.


          • #6
            I had Performance Test 8 fail in the same manor. I downloaded it a little over a week ago and implemented the same script that loops continuously. In my case, it failed after the start of the second iteration.

            The system is a core i7-3615QE with a QM77 PCH. It was a fresh install of windows 7 64-bit with SP1 and all the latest updates. The drivers were all updated as far as the vendors and windows update was concerned.

            David, I was in contact with passmark support over the SMBus locking up just by launching BurninTest Pro 7 and it querying the specifics about the system. I provided debug logs using bit.exe -u


            • #7
              A bug was discovered that caused any scripting involving the DirectCompute test to freeze after the first stage of the DirectCompute test. The latest build,
              should have fixed this.

              Are you sure you actually made it through to the second iteration and that it didn't freeze after the DirectCompute test? Either way, can you give the new build a go and if there is still a problem can you email us the exact script file you are using.


              • #8

                I am not sure if it locked up on the Direct Compute test but I will check on that.
                In the mean time, I have emailed the script we were using on PT7 and PT8.

                I have prepared a video of what we observe when the 3D test fails - it is ~43MB
                You can find it at the following link:

                Please use VideoLAN or K-lite mega codec pack to view.