MFX_ERR_GPU_HANG when using several decoding/encoding applications

[OTorg]

This topic was born during splitting

https://software.intel.com/en-us/forums/intel-media-sdk/topic/827645#comment-1946714

into two different themes.

[Mark_L_Intel1]

Thanks so much,

I can reproduce it with following error message, could you confirm? My environment is skull Cannon NUC and Windows 10, driver 6373.

Encoding Sample Version 8.4.27.378

Input file format       NV12
Output video            AVC
Source picture:
        Resolution      720x576
        Crop X,Y,W,H    0,0,720,576
Destination picture:
        Resolution      720x576
        Crop X,Y,W,H    0,0,720,576
Frame rate      25.00
Bit rate(Kbps)  3000
Gop size        28
Ref dist        4
Ref number      4
Idr Interval    0
Target usage    speed
Memory type     system
Media SDK impl          hw
Media SDK version       1.27

Processing started
Frame number: 286800
[ERROR], sts=MFX_ERR_GPU_HANG(-21), CEncTaskPool::SynchronizeFirstTask, SyncOperation failed at src\pipeline_encode.cpp:157

[ERROR], sts=MFX_ERR_GPU_HANG(-21), CEncodingPipeline::GetFreeTask, m_TaskPool.SynchronizeFirstTask failed at src\pipeline_encode.cpp:1748

[ERROR], sts=MFX_ERR_GPU_HANG(-21), CEncodingPipeline::Run, m_pmfxENC->EncodeFrameAsync failed at src\pipeline_encode.cpp:1961

[ERROR], sts=MFX_ERR_GPU_HANG(-21), wmain, pPipeline->Run failed at src\sample_encode.cpp:1301
Frame number: 286872
Encoding fps: 227

C:\Users\test\Documents\Intel® Media SDK 2018 R2 - Media Samples 8.4.27.378\_bin\x64>

I didn't get it at the first run(about 20minutes), but I got it at the second run(about 30minutes) but I don't know when it happened since I saw it when I came back to screen.

It looks like 4 console spawned during the run time(see the attached screen capture), at the start, "Console 1" only spawned 2 consoles, and I think "console 4" and "console 5" were spawned after "console 2" and "console 3" crashed. When I came back, "console 5" was running but "console 4" was hanged.

I have submitted the bug and will keep you updated.

Mark

[OTorg]

Hi Mark,

Liu, Mark (Intel) wrote:
 I can reproduce it with following error message, could you confirm? 

Yes, I confirm that this is the bug I was talking about.

[OTorg]

Hi!

Is there any news on MFX_ERR_GPU_HANG fixing?

[Mark_L_Intel1]

Not yet, I will ask them to do a brief check. Could you check it again next Wednesday?

Mark

[OTorg]

Liu, Mark (Intel) wrote:

Could you check it again next Wednesday?

Yes, I can. Waiting for instructions...

[Mark_L_Intel1]

Hi Alek,

Sorry for the late response, I discuss with dev team about this issue and the original issue you reported. They start working on it now.

They will first check if these are bugs or not and if they are, they are going to fix them.

Mark

[Mark_L_Intel1]

Hi Alek,

My apology, I might miss responding to this issue, dev team had a suggestion:

They want to see if you can try MSS 2018-R2.3 and see if this release can solve the issue.

Again my apology for not updating the solution, can you make a try on this release?

Mark Liu

[OTorg]

Liu, Mark (Intel) wrote:
 They want to see if you can try MSS 2018-R2.3 and see if this release can solve the issue. 

Hi Mark,

Where can I find windows binaries of that release?

[Mark_L_Intel1]

Sorry Alek,

This is my mistake, my post was not relevant to your issue. 

For your issue, the latest update is, we started the investigation, it is at early stage and I will give you update when we found something.

Mark

[BMart1]

Hi,

Our Windows application that does several simultaneous real-time encodings also suffers from MFX_ERR_GPU_HANG, even with the latest driver version 27.20.100.8336. Any news regarding a fix?

Thanks,
Bruno

[Luis_at_Intel]

Hi @BMart1,

This issue is still under investigation, we will provide an update when available. Thank you for your patience.

Regards,

Luis 

Ref. 59912

[Mark_L_Intel1]

Hi Bruno,

This issue was reported and tracked very well. Based on the record, we had a discussion on it about June, we decided to delay it since the other urgent project and resume when our resources are available.

Is your GPU hang issue is the same as Alek? If it is different, you can submit a different request either on Forum or online support center.

Mark

[BMart1]

Hi Mark,

It's hard to tell from the outside whether the issue is identical or not. Our app certainly is different. We do multiple real time encodes and get the same -21 error.

Bruno

[Mark_L_Intel1]

Hi Bruno,

Thanks for the comment.

Understood your situation. It should take time to create a reproducer, this would always be preferred so if you happen to have it, please post. Otherwise, I will focus on fixing this report.

I will add your comment to the report.

Mark

[Pamela_H_Intel]

Apologies for the delays. Media SDK functionality has been superseded by oneVPL, FFmpeg and GStreamer.

For more information see the oneVPL Landing Page: https://www.intel.com/content/www/us/en/developer/tools/oneapi/onevpl.html where you will find, among other information

oneVPL version 2.x is an API that is the successor to Intel® Media SDK. The Intel Media SDK project is no longer active and will not be updated. Packages and downloads are still available on GitHub*. For continued support and access to new features, Intel Media SDK users are encouraged to move to oneVPL as soon as possible.

·       Upgrade from Intel Media SDK to oneVPL

·       Intel Media SDK Discontinuation Notice

·       Download from GitHub

Let us know if you have any questions about oneVPL.