Hello,

Thanks for letting us know about the issue that you have found.

As you indicated, it sounds like this issue is isolated to the specific game/application that you are analyzing, since other DX11 apps don't have this issue (including those with the compute shader).

By the way, you mentioned that a colleague is able to run ok with the same app with another version of Intel GPA -- do you know what graphics device she or he is using? I'm trying to determine whether this is due to a specific version of Intel GPA (and it could be due to the file creation or file playback), or whether it's partly due to the graphics device that you are using.

Also, having access to a copy of the frame capture file would help, but your answer to a couple of the other questions might help us get more info up-front before needing the file.

Regards,

Neal

Hello,

Thanks for the additional information. The part about your app running "clean" (with break on error enabled) is important, as this eliminates a key item that we've seen before (and yes, we are working to make the program more robust).

I realize that it's probably not easy for us to access your frame capture file.

Can you wait a couple of days? I may have another version of GPA that I can provide you for testing to see if this fixes the problem.

Regards,

Neal

Hello,

I've got a new version for you to check out -- goto the Intel GPA Home Page, then under the "downloads" section look for "2012 R1".

Please let me know whether this version helps.

Regards,

Neal

Hello,

I'm checking with the dev team for some more suggestions on how to proceed with tracking down the issue.

In the meantime, you mentioned that a colleague of yours is able to run your app ok on Intel GPA 4.0 (including both capturing the file and viewing it in Intel GPA Frame Analyzer). I believe that you said that when you run your app on your machine you can capture the file ok but not view it in Intel GPA Frame Analyzer (issues with textures and render target previews on 4.0, but a crash on 4.3 or 12.1).

Do you know whether your colleague has run your app OK on the 4.3 or 12.1 versions of Intel GPA? I guess I'm wondering whether the issue is somehow related to your specific device and/or driver version.

Regards,

Neal

I have the same error, in my managed project (I'm using SharpDX DirectX11 wrapper). I managed to track down "THE" line that make this error occurs for me : context.CopyResource(_engine.BackBufferTex, _solidBackBuffer); Where I simply do a copy of the BackBuffer texture resource into another similar texture. Without this line, the error is not raised. Don't know If it can help. Regard.

Hello,

Thanks for the info.

Could you please include your configuration information by right-clicking the Intel GPA Monitor icon in the notification tray and selecting "About..."?

Also, are you running with "break on error"? If so, could you run without this option and let us know if you see a difference?

thanks!

Neal

Hello,

We've not been able to reproduce this issue internally... a frame capture file from you would be very helpful!

Regards,

Neal

For sure, here is the Info about my system first (this machine is really low end one) :
Windows 7, 32-bit DEP enabled Num Processors: 2 Memory: 3583MB System BIOS: American Megatrends Inc. 0401 (07/14/2009) Video BIOS: Version 62.98.42.00.00 (02/04/09) (Display driver information not available) GPA install directory: C:\Program Files\Intel\GPA\2012 R1\ GPA version: 12.1.166792 Current user is in Administrators group: YES Current GPA 2012 R1 (12.1.166792)

Windows 7, 32-bit DEP enabledNum Processors: 2Memory: 3583MBSystem BIOS: American Megatrends Inc. 0401 (07/14/2009)Video BIOS: Version 62.98.42.00.00 (02/04/09)(Display driver information not available)GPA install directory: C:\Program Files\Intel\GPA\2012 R1\GPA version: 12.1.166792Current user is in Administrators group: YESCurrent GPA 2012 R1 (12.1.166792) I have uploaded the frame causing probleme here :http://www.sendspace.com/file/z2rcz5 Regards.

Hello,

Thanks for the file... I've copied it for analysis by the development team. I'll let you know as soon as I hear back from them.

Regards,

Neal

Tx you. One remark, maybe important. I tried to replay this from another computer, without any problem. The "big" difference I see, is that the problem is occuring from a Dual-monitor configuration. Will try on another dual screen configuratin, and keep you informed.

Hello,

The Development Team has been able to duplicate this problem and provide a workaround.

The error seems to occur only when you are using the D3D control panel to enable "break settings" for errors in your code. If you disable this feature, Intel GPA should properly create, read, and process the frame capture file.

Therefore, as a quick fix please disable any "break" settings, and let me know if this workaround resolves your issue.

Regards,

Neal

Hello, I'm sorry to tell you that the D3D control panel, "Break settings" is empty. And that this error is still occuring when the DirectX device is created without "Debug flag". It clearly seems to be linked to computer, as the same program can be debugged easily on another hardware. Regards.

Hello,

It appears that we may have multiple issues here.

On one of our internal systems we did see the same error with the capture file you provided, and it ended up being due to the D3D control panel settings. It's been fixed internally, and is planned for the next release.

We also saw that this bug could occur when the NVIDIA driver is reloaded by Windows 7* -- this occured when using trivial shaders with some versions of the NVIDIA driver, and only on some h/w configurations, and we are actively looking at this situation as well. This may be why the workaround we suggested didn't work (that is, don't use "break on error").

You also mentioned that you were going to test on another dual-monitor configuration -- did this configuration work ok?

And one final question -- how critical is a fix for this bug to you? Is the workaround of using another system sufficient for you, or is this seriously affecting your analysis/optimization tasks?

Also,

Hello,

As I mentioned the development team has been doing additional testing of this issue, and we have found what we believe is the root cause of this problem.

Our testing shows that the latest NVidia driver (296.10) causes memory leaks if a D3D
application uses UpdateSubresource D3D call in some situations. The Intel GPA playback
library uses this call, and thats why we see crash in playback when opening this frame in Intel GPA Frame Analyzer.

We do not see this problem with the NVidia 266.58 driver -- this seems to work ok and do not have the problem with Intel GPA Frame Analyzer.

As we create a workaround for this in our code, would you be able to test with the 266 driver, and see if the problem still occurs? Your testing will help us verify that we are on the right track in getting Intel GPA to work correctly.

thanks!

Neal