OpenCL: hanging process or crash when changing the tolerance on selecting sampled color dialog (Beta - 1.9.0.734)

[gkalab]

Beta 1.9.0.734 crashes or hangs the process when you increase the tolerance when selecting a sampled color. This only happens when OpenCL is enabled.

Steps to reproduce:

Create a new document
Create a new empty pixel layer
Select sampled color
Increase the tolerance

[Chris B]

Hey gkalab,

I'm struggling to reproduce this on two machines here. If you're crashing to the desktop, can you see if you have a crash report in:
C:\Users\yourusername\AppData\Roaming\Affinity\Photo\1.0 (Beta)\CrashReports\reports

[Joachim_L]

Happens for me too, first time AP was not responding (had to terminate) and second it was directly crashing to desktop. But there are no crash reports. To get the crash I have to move the slider several times forth and back.

PS: Having acceleration activated there is a one second lag when add an empty pixel layer.

[gkalab]

There are no crash reports in C:\Users\<usernam>\AppData\Roaming\Affinity\Photo\1.0 (Beta)\CrashReports. Is there any other log file I can send?

I can still reproduce the issue with the latest beta (1.9.0.767). It seems it is now harder to reproduce (don't know why): now I need to move the slider several times forth and back as @Joachim_L mentioned. Most of the time the application just hangs and I have to kill it, sometimes it crashes.

I made sure I installed the latest graphic drivers from Intel:

Processor: Intel(R) Core(TM) i5-8400 CPU @ 2.80GHz, Graphics card: on-board Intel(R) UHD Graphics 630UHD

[Joachim_L]

23 minutes ago, gkalab said:

I made sure I installed the latest graphic drivers from Intel:

As we have the same graphics card, what is your actual driver version? Mine, 26.20.100.7757, is working now without problems ... in this case.

[gkalab]

7 minutes ago, Joachim_L said:

As we have the same graphics card, what is your actual driver version? Mine, 26.20.100.7757, is working now without problems ... in this case.

Mine is now 26.20.100.8783.

[Chris B]

I still cannot get this to happen and I've been trying for a solid 20+ mins non-stop!

Can one of you kindly save an .afphoto file, with its History and attach it please? 

 

[gkalab]

4 minutes ago, Chris B said:

I still cannot get this to happen and I've been trying for a solid 20+ mins non-stop!

Can one of you kindly save an .afphoto file, with its History and attach it please? 

 

Well - there is no history except creating a new pixel layer 🙂 - but I saved a file for you.

It takes me about half a minute dragging the slider left and right to make the application hang.

Again the steps I took to reproduce the issue:

- Create a new document (I chose 800x600)
- Create a new empty pixel layer
- This time I saved the document to the attached file.
- Choose from the menu: Select sampled color
- Drag the slider for one minute or so.

This time there's also an autosave file (attached).

Thank you.

example.afphoto 20201001145628.autosave

[gkalab]

Still the same on

Beta - 1.9.0.791

Takes 10 to 30 seconds dragging the slider left and right...

[Chris B]

Have you checked the Task Manager to see if your CPU is hitting 100%? 

I'm not sure if it's potentially draining resources and Windows is killing the app or if there's an actual bug in the app but I just cannot crash it.

Is it 30 seconds of constant dragging without release or lots of little drags for 30 seconds?

[gkalab]

6 hours ago, Chris B said:

Have you checked the Task Manager to see if your CPU is hitting 100%? 

I'm not sure if it's potentially draining resources and Windows is killing the app or if there's an actual bug in the app but I just cannot crash it.

Is it 30 seconds of constant dragging without release or lots of little drags for 30 seconds?

It's constant dragging without release. Most of the time the program is just hanging - see thread list. Yes, it seems one core is a little busy. BTW, while dragging it's always the clReleaseGlSharedEvent thread that's on the top.

Attached are the CPU graph, GPU graph and thread list when the program hangs. The GPU spike is from when I was constantly dragging.

[Chris B]

I've just tried again with and without my GPU and it doesn't crash, however I am seeing some strange render/tiling issue on a blank doc with just a pixel layer. 

[gkalab]

23 minutes ago, Chris B said:

I've just tried again with and without my GPU and it doesn't crash, however I am seeing some strange render/tiling issue on a blank doc with just a pixel layer. 

Thank you for trying 🙂

I'm wondering: are you trying to reproduce this issue on an Intel 630UHD graphics card? If not, maybe that's the reason you are failing to reproduce the issue?

Also, is there really no log file the applicaction is writing in the debug version?

[Chris B]

19 hours ago, gkalab said:

I'm wondering: are you trying to reproduce this issue on an Intel 630UHD graphics card? If not, maybe that's the reason you are failing to reproduce the issue?

 

Yes—I have an Alienware laptop that has a Intel UHD Graphics 630 and an RTX 2060. With OpenCL enabled and the Renderer set to the 630, I get very high GPU usage and 50% RAM usage but no crash. The fans start blowing very aggressively but I've dragged for 2 minutes straight and nothing. 

The only place you're going to find a log is in:
C:\Users\yourusername\AppData\Roaming\Affinity\Photo\1.0 (Beta)\CrashReports\reports

The Windows Event Viewer might tell us something if you don't have a report.

[gkalab]

On 10/27/2020 at 12:10 PM, Chris B said:

Yes—I have an Alienware laptop that has a Intel UHD Graphics 630 and an RTX 2060. With OpenCL enabled and the Renderer set to the 630, I get very high GPU usage and 50% RAM usage but no crash. The fans start blowing very aggressively but I've dragged for 2 minutes straight and nothing. 

The only place you're going to find a log is in:
C:\Users\yourusername\AppData\Roaming\Affinity\Photo\1.0 (Beta)\CrashReports\reports

The Windows Event Viewer might tell us something if you don't have a report.

The reports folder is empty. Maybe because the application hangs and does not crash any more (since the first beta or so)?

Here are the Windows Event Viewer entries - unfortunately I only have a German Windows 10 version:

1) System log Warning 28.10.2020 21:38:41: Der Anzeigetreiber "igfx" reagiert nicht mehr und wurde wiederhergestellt.

2) Application Info 28.10.2020 21:38:43:

Fehlerbucket , Typ 0
Ereignisname: LiveKernelEvent
Antwort: Nicht verfügbar
CAB-Datei-ID: 0

Problemsignatur:
P1: 141
P2: ffff8905d7cab010
P3: fffff8060e524cb0
P4: 0
P5: 2ce4
P6: 10_0_19041
P7: 0_0
P8: 256_1
P9:
P10:

Angefügte Dateien:
\\?\C:\WINDOWS\LiveKernelReports\WATCHDOG\WATCHDOG-20201028-2138.dmp
\\?\C:\WINDOWS\TEMP\WER-187532390-0.sysdata.xml
\\?\C:\ProgramData\Microsoft\Windows\WER\Temp\WER8F43.tmp.WERInternalMetadata.xml
\\?\C:\ProgramData\Microsoft\Windows\WER\Temp\WER8F44.tmp.xml
\\?\C:\ProgramData\Microsoft\Windows\WER\Temp\WER8F53.tmp.csv
\\?\C:\ProgramData\Microsoft\Windows\WER\Temp\WER8F64.tmp.txt

Diese Dateien befinden sich möglicherweise hier:
\\?\C:\ProgramData\Microsoft\Windows\WER\ReportQueue\Kernel_141_85ea1efb85582c374233b5b5eb7df9b465ef758_00000000_b93c5b0b-f090-4298-9f4b-b958026937d8

Analysesymbol:
Es wird erneut nach einer Lösung gesucht: 0
Berichts-ID: b93c5b0b-f090-4298-9f4b-b958026937d8
Berichtstatus: 4
Bucket mit Hash:
CAB-Datei-Guid: 0

3) Application Info 28.10.2020 21:38:43:

Fehlerbucket , Typ 0
Ereignisname: LiveKernelEvent
Antwort: Nicht verfügbar
CAB-Datei-ID: 0

Problemsignatur:
P1: 141
P2: ffff8905d7cab010
P3: fffff8060e524cb0
P4: 0
P5: 2ce4
P6: 10_0_19041
P7: 0_0
P8: 256_1
P9:
P10:

Angefügte Dateien:
\\?\C:\WINDOWS\LiveKernelReports\WATCHDOG\WATCHDOG-20201028-2138.dmp
\\?\C:\WINDOWS\TEMP\WER-187532390-0.sysdata.xml
\\?\C:\ProgramData\Microsoft\Windows\WER\Temp\WER8F43.tmp.WERInternalMetadata.xml
\\?\C:\ProgramData\Microsoft\Windows\WER\Temp\WER8F44.tmp.xml
\\?\C:\ProgramData\Microsoft\Windows\WER\Temp\WER8F53.tmp.csv
\\?\C:\ProgramData\Microsoft\Windows\WER\Temp\WER8F64.tmp.txt

Diese Dateien befinden sich möglicherweise hier:
\\?\C:\ProgramData\Microsoft\Windows\WER\ReportArchive\Kernel_141_85ea1efb85582c374233b5b5eb7df9b465ef758_00000000_cab_b93c5b0b-f090-4298-9f4b-b958026937d8

Analysesymbol:
Es wird erneut nach einer Lösung gesucht: 0
Berichts-ID: b93c5b0b-f090-4298-9f4b-b958026937d8
Berichtstatus: 2049
Bucket mit Hash:
CAB-Datei-Guid: 0

--------------------

After closing the hanging app:

4) Application Error 28.10.2020 21:39:39

Das Programm Photo.exe Version 1.9.0.820 hat die Interaktion mit Windows beendet und wurde geschlossen. Überprüfen Sie den Problemverlauf in der Systemsteuerung "Sicherheit und Wartung", um nach weiteren Informationen zum Problem zu suchen.
 Prozess-ID: 2ce4
 Startzeit: 01d6ad6a3e17f508
 Beendigungszeit: 10
 Anwendungspfad: C:\Program Files\Affinity\Photo Customer Beta\Photo.exe
 Bericht-ID: 64f2d917-e691-47b2-af95-879c9c984797
 Vollständiger Name des fehlerhaften Pakets:
 Relative Anwendungs-ID des fehlerhaften Pakets:
 Absturztyp: Unknown

5) Application Info 28.10.2020 21:39:39

Fehlerbucket , Typ 0
Ereignisname: AppHangB1
Antwort: Nicht verfügbar
CAB-Datei-ID: 0

Problemsignatur:
P1: Photo.exe
P2: 1.9.0.820
P3: 5f95f25f
P4: 008a
P5: 134217728
P6:
P7:
P8:
P9:
P10:

Angefügte Dateien:
\\?\C:\ProgramData\Microsoft\Windows\WER\Temp\WER5514.tmp.WERInternalMetadata.xml
\\?\C:\ProgramData\Microsoft\Windows\WER\Temp\WER5534.tmp.xml
\\?\C:\ProgramData\Microsoft\Windows\WER\Temp\WER5535.tmp.csv
\\?\C:\ProgramData\Microsoft\Windows\WER\Temp\WER5546.tmp.txt
\\?\C:\Users\Gerhard\AppData\Local\Temp\WER6840.tmp.appcompat.txt

Diese Dateien befinden sich möglicherweise hier:
\\?\C:\ProgramData\Microsoft\Windows\WER\ReportArchive\AppHang_Photo.exe_78b7bbace0373c4ca77c2cd0f8ae6f4635367c47_a854876b_7852583e-4b3c-4d98-b16c-e76cd4dc5a25

Analysesymbol:
Es wird erneut nach einer Lösung gesucht: 0
Berichts-ID: 64f2d917-e691-47b2-af95-879c9c984797
Berichtstatus: 1
Bucket mit Hash:
CAB-Datei-Guid: 0

------------------

The folder C:\ProgramData\Microsoft\Windows\WER\ReportArchive\AppHang_Photo.exe_78b7bbace0373c4ca77c2cd0f8ae6f4635367c47_a854876b_7852583e-4b3c-4d98-b16c-e76cd4dc5a25 contained a Report.wer file, which I have attached.

 

Report.wer

[gkalab]

I played around a bit and installed oclgrind, which is a "virtual OpenCL device simulator" and it is displayed as an additional OpenCL Hardware in Photo Beta.

Together with oclgrind I was unable to reproduce the issue - I got no hang.

Then I've uninstalled oclgrind and I can reproduce the hang on an empty pixel layer (it is more difficult to reproduce with actual image data).

So maybe the UHD630 has to be the only graphics card installed to reproduce the issue?

On a side note what's also interesting is that I don't get the 2 second lag when adding the first pixel layer on a new document after starting photo when oclgrind is installed. I do get the lag adding the first pixel layer after starting photo every time when oclgrind is not installed.

[Chris B]

15 hours ago, gkalab said:

So maybe the UHD630 has to be the only graphics card installed to reproduce the issue?

 

Potentially yes. That's an interesting bit of detective work—thank you. I will review the info (I managed to translate it easily).