1.7.427 final denoise crash

Bugs, glitches and unexpected behaviour.
Post Reply
clouzot
Posts: 9
Joined: Thu Jul 23, 2020 7:56 pm

1.7.427 final denoise crash

Post by clouzot »

Hi Ivo,

Using 1.7.427a, macOS 10.15.7

The usual workflow went OK until I clicked on the final denoise (track), where Startools crashed immediately.

Attached is the crash log
StarTools_2020-10-19-103213.crash.zip
(17.86 KiB) Downloaded 10 times
Also attached is startools.log, but it stops after the Color module work and has no further entries
StarTools.log.zip
(173.59 KiB) Downloaded 9 times
Cheers,

Ben
User avatar
admin
Site Admin
Posts: 2501
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

Re: 1.7.427 final denoise crash

Post by admin »

Hi Ben,

Thank you for this helpful report. I haven't heard from anyone else having this issue (so far)...
It appears from the very helpful log that it's the GPU (Intel HD 5000) driver that is crashing when trying to hand back a processed result to the CPU.

Can you tell me what the last user interaction was before the crash? (e.g. which button you pressed exactly)

Thank you!

EDIT: Also, can you confirm for me you are not mixing and matching the application with an out-of-date 'resources' file? Thanks!
Ivo Jager
StarTools creator and astronomy enthusiast
clouzot
Posts: 9
Joined: Thu Jul 23, 2020 7:56 pm

Re: 1.7.427 final denoise crash

Post by clouzot »

Hi Ivo,

oddly enough my MacBook Pro has two GPUs, the crappy built-in Intel Iris Pro, and a (crappy now, but decent when I bought it) nVidia GT750M.

I can see that Startools is indeed using the nVidia one with most modules (see the effect of Wipe, for instance) so I wonder why the Intel GPU crashed in the first place. As far as I know, the Intel GPU is only used for the compositing window management (what macOS calls Mission Control).
Screenshot 2020-10-20 at 09.41.59.jpg
Screenshot 2020-10-20 at 09.41.59.jpg (359.75 KiB) Viewed 199 times
The last user interaction was, just after using the Color module, to click on Track/NR, where Startools closed with no further notice. For some reason, macOS didn't display the crash notice window so I had to dig the log.

I just checked and I'm indeed using the correct resource file (the one that came with the zipped 1.7 "distro"). The only thing I did was to copy the 1.6 license file to that folder.

Thanks again for that marvelous piece of software. The GPU-enabled version speeds up thing enormously!
User avatar
admin
Site Admin
Posts: 2501
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

Re: 1.7.427 final denoise crash

Post by admin »

Totally my bad - I misread the log. It's indeed the GT 750M driver throwing the error.

What puzzles me is that there should be no GPU activity at all when you click Track/NR; an interstitial popup should come up where, depending on your choice, new modules are launched that do tax the GPU. :confusion-shrug:

Can you replicate this behavior consistently?
Ivo Jager
StarTools creator and astronomy enthusiast
clouzot
Posts: 9
Joined: Thu Jul 23, 2020 7:56 pm

Re: 1.7.427 final denoise crash

Post by clouzot »

admin wrote: Wed Oct 21, 2020 12:00 am Can you replicate this behavior consistently?
Unfortunately not. It would be too easy :lol:
I've processed a couple FITS since then, with no crash whatsoever. So I guess it was a random crash, maybe due to the nVidia driver itself (hard to say).
User avatar
admin
Site Admin
Posts: 2501
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

Re: 1.7.427 final denoise crash

Post by admin »

clouzot wrote: Wed Oct 21, 2020 6:46 am
admin wrote: Wed Oct 21, 2020 12:00 am Can you replicate this behavior consistently?
Unfortunately not. It would be too easy :lol:
I've processed a couple FITS since then, with no crash whatsoever. So I guess it was a random crash, maybe due to the nVidia driver itself (hard to say).
Ok, please do let me know if this is something that happens often or can be replicated in some way!
Ivo Jager
StarTools creator and astronomy enthusiast
Post Reply