Denoise(profiled) Crashes 3.4

Got it! :+1:

You mean, you edited the old xmp file but with darktable fresh started (the old appdata darktable folder renamed)?

A 1000x typo, should have been 32G. :blush:

1 Like

Normally when an error condition occurs you should have a graceful handling of it, e.g., error message, etc. But it should not crash. Hard to know if in this case darktable is able to detect the problem before things go totally haywire.

This is what I did

  1. Renamed darktable folder in appdata, renamed the original xmp. Opened darktable, imported same image as a new one and edited - no problems
  2. Deleted the daarktable folder in appdata, copied original xmp file back, restarted darktable so now a fresh darktable but with original xmp - couldn’t get darktable to recognise the xmp file i.e. image not showing with correct edits - gave up
  3. Put original darktable folder back in appdata and left original xmp file. Opened darktable (now in original state) edited the same problem image - no problems (this is the original set up that used to crash)

As I said I’ll keep an eye on it and try and diagnose it better

1 Like

Thanks for that good info concerning Windows. I have 32gb on the Win10 computer I have 3.4.0 installed.

So, if you keep the whole original setup (config folder and xmp), remove the database entry from darktable (in lightroom) and then import it again (with the original xmp present), I wonder if it shows the symptoms of #2.

As an update, forget I ever raised the topic.
Processed about 30 photos over the last couple of days and DT stable as ever.
Not sure what it was, but more than likely an instability in my system rather than any issue with DT
Sorry for wasting peoples time and thanks for all your help

1 Like

I experienced crashes on Linux, including in profiled denoise, after a lot of edits, but haven’t created a backtrace yet (gdb wasn’t installed). In that sense, you sure have not wasted my time, and I’ll see if I can create reproduceable scenario. But then I have experienced other issues (likely in the opencl code) and it could all be related to my pieced-together driver setup …