1.7.424 alpha with new Wipe module

General discussion about StarTools.
georgemsimon
Posts: 1
Joined: Sat Sep 12, 2020 5:10 pm

Re: 1.7.424 alpha with new Wipe module

Post by georgemsimon »

Ivo,

The latest release is a big improvement over the previous one! With the previous one, I consistently experienced crashes when using Denoise 2. I just processed a mediocre data set with the latest release, and there were no crashes. In addition, it might be my imagination, but the latest release seemed to be a bit faster on all modules.

Many thanks for your continued work on this. It is truly amazing!

George Simon
happy-kat
Posts: 372
Joined: Sun Feb 01, 2015 11:31 am

Re: 1.7.424 alpha with new Wipe module

Post by happy-kat »

Have used this new release 1.7.427 GPU tonight, in general it did feel responsive as per previous comment.
First process crashed shut in Wipe.
Second process got past wipe and crashed shut when adjusting a % in Autodev.
Third process attached.
The new Wipe presets are great though this image responds best to 95% aggressive on adaptive vertical bias.
The Autodev did pull more nebula through on this process approach but at the cost of the quality of the applied Wipe which had been successful.
Autosave v3.1.jpg
Autosave v3.1.jpg (196.38 KiB) Viewed 4037 times
just trying another image, NeoWise that I have not yet processed, Wipe crashed shut on me. it is a very challenging image as had static trees in foreground so was using a mask to avoid the comet, comet tails and trees plus taken before astro dark. I will try again tomorrow.
User avatar
admin
Site Admin
Posts: 3367
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

Re: 1.7.424 alpha with new Wipe module

Post by admin »

georgemsimon wrote: Mon Oct 12, 2020 6:51 pm The latest release is a big improvement over the previous one!
Fantastic to hear!
happy-kat wrote: Mon Oct 12, 2020 7:55 pm First process crashed shut in Wipe.
Second process got past wipe and crashed shut when adjusting a % in Autodev.
Wipe crashed shut on me
Hmmm... this is starting to worry me as the instabilities seem to exist across the application in your case.
Crashing in AutoDev really should not be happening.

You indicated your have an i5 8600 and a GT 730 2GB.
You could try using your i5-8600's iGPU instead to see if it is more stable. There is a good chance that it may actually also be faster than your GT 730, which is rather underpowered.

To make StarTools use your iGPU, create a file named opencldeviceindex.cfg in the StarTools folder (e.g. where the executables, 'resources', 'CHANGELOG', 'StarTools.log', etc. files are located). In that file, put the character 0 (zero). That's all.

This will force StarTools to use the "0th" device it finds, which should be your iGPU.
Ivo Jager
StarTools creator and astronomy enthusiast
happy-kat
Posts: 372
Joined: Sun Feb 01, 2015 11:31 am

Re: 1.7.424 alpha with new Wipe module

Post by happy-kat »

Thank you Ivo, I will try that. I use the graphics card still as the PC onboard chip capabilities has poor range in colour calibration, I need the graphics card to see all the finer tones on the colour chart you shared some time ago. I will try the startools switch txt.
happy-kat
Posts: 372
Joined: Sun Feb 01, 2015 11:31 am

Re: 1.7.424 alpha with new Wipe module

Post by happy-kat »

unfortunately the GPU version with that file I created has crashed 3 times in 3 uses. the modules that crashed StarTools shut were
Wipe
Autodev
Colour
I was using my comet stack that is very challenging in the data it presents, I suspect I may have to process it by hand.
Perhaps tomorrow I will try the dataset I had shared to see how that behaves.
I think my onboard graphics capability may be a lower spec then assumed.
User avatar
admin
Site Admin
Posts: 3367
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

Re: 1.7.424 alpha with new Wipe module

Post by admin »

happy-kat wrote: Tue Oct 13, 2020 9:11 pm unfortunately the GPU version with that file I created has crashed 3 times in 3 uses. the modules that crashed StarTools shut were
Wipe
Autodev
Colour
I was using my comet stack that is very challenging in the data it presents, I suspect I may have to process it by hand.
Perhaps tomorrow I will try the dataset I had shared to see how that behaves.
I think my onboard graphics capability may be a lower spec then assumed.
Hmmm...
Can you confirm for me that StarTools indeed has started to use your iGPU (according to the info in the About dialog or the splash screen) instead of the GT 730? The crashes happening all over the place (e.g in various unrelated parts of the code) are quite worrying, and it is not something I can replicate with the systems and GPUs I have here (I've started to amass a few...).
Ivo Jager
StarTools creator and astronomy enthusiast
happy-kat
Posts: 372
Joined: Sun Feb 01, 2015 11:31 am

Re: 1.7.424 alpha with new Wipe module

Post by happy-kat »

Looks to not be doing the switch
Startools is reporting the GT 730 card in About
The onboard graphics is an intel(R) UHD Graphics 630
The PC is an i5 with 6 cores
happy-kat
Posts: 372
Joined: Sun Feb 01, 2015 11:31 am

Re: 1.7.424 alpha with new Wipe module

Post by happy-kat »

Hi Ivo, the switch you gave to try above did not switch for me
User avatar
admin
Site Admin
Posts: 3367
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

Re: 1.7.424 alpha with new Wipe module

Post by admin »

happy-kat wrote: Fri Oct 16, 2020 4:16 pm Hi Ivo, the switch you gave to try above did not switch for me
I uploaded this 2-byte file;
https://download.startools.org/opencldeviceindex.cfg
Could you save this into the StarTools folder (e.g. where the executables, 'resources', 'CHANGELOG', 'StarTools.log', etc. files are located) and see if that works?
Ivo Jager
StarTools creator and astronomy enthusiast
happy-kat
Posts: 372
Joined: Sun Feb 01, 2015 11:31 am

Re: 1.7.424 alpha with new Wipe module

Post by happy-kat »

Thanks Ivo
I tried your version of that file, no change it made no difference the GT card is still being used by Startools.
Though your file looked identical to the one I had manually created, it also just had a 0 inside.
Post Reply