1.7.430 alpha with new Super Structure module (Life replacement)

[admin]

there are no longer any streaking in the images.

Thank you so much for confirming this. This has helped me narrow down the problem; it is either a GPU memory allocation issue or - much more likely - a GPU timeout issue (e.g. GPU taking too long to process to the operating system's liking).

I am hopeful the fixes in 1.7.440 have addressed this, Do let me know if the issue persists!

[Stefan B]

Thanks Stefan. It's the information displayed when you click 'About' in the home screen;

Hi Ivo,

sorry for the late reply. You'll find the about dialog attached. 1.7.440 works for me so far. Thanks!

Best regards
Stefan

[cytan299]

there are no longer any streaking in the images.

Thank you so much for confirming this. This has helped me narrow down the problem; it is either a GPU memory allocation issue or - much more likely - a GPU timeout issue (e.g. GPU taking too long to process to the operating system's liking).

I am hopeful the fixes in 1.7.440 have addressed this, Do let me know if the issue persists!

Hi Ivo,
Problem persists with 1.7.440. Here's the results.
dimsmall (looks like no streaks, but I see it when I zoom in, zoomed out version can be seen in my next post):

dimsmall_zoom.jpg (48.8 KiB) Viewed 7139 times

airy:

airy.jpg (351.53 KiB) Viewed 7139 times

isolate:

isolate.jpg (432.9 KiB) Viewed 7139 times

cytan

[cytan299]

there are no longer any streaking in the images.

Thank you so much for confirming this. This has helped me narrow down the problem; it is either a GPU memory allocation issue or - much more likely - a GPU timeout issue (e.g. GPU taking too long to process to the operating system's liking).

I am hopeful the fixes in 1.7.440 have addressed this, Do let me know if the issue persists!

And heres dimsmall zoomed out:

dimsmall.jpg (424.25 KiB) Viewed 7139 times

If you want the fits file to play with, you can download it here:

https://1drv.ms/u/s!ApsRFMpkd9cwgRZuTBU ... M?e=bhcPp4

cytan

[admin]

Problem persists with 1.7.440. Here's the results.

I'm afraid we're quickly nearing the end of the road for your particular machine.

Are you absolutely, 100% sure you are running the latest version (1.7.440alpha, including its 'resources' file)?
Are you absolutely, 100% sure your RX 5500M is being used and not, say, your iGPU? (see About dialog)

Thank you!

[cytan299]

Problem persists with 1.7.440. Here's the results.

I'm afraid we're quickly nearing the end of the road for your particular machine.

Are you absolutely, 100% sure you are running the latest version (1.7.440alpha, including its 'resources' file)?
Are you absolutely, 100% sure your RX 5500M is being used and not, say, your iGPU? (see About dialog)

Thank you!

Here's the screenshot from StarTools:

track.jpg (46.9 KiB) Viewed 7114 times

And here's the checksum of the resources file and size:

Code: Select all

/Applications/Startools@belle% cksum resources
1691887530 1887771 resources

/Applications/Startools@belle% ls -l resources
-rwxrwxrwx@ 1 cytan  staff  1887771 Nov  6 09:54 resources*

Is this what you'd expect? I double checked that the resources file has the same checksums as that in the zip distribution.

Do you see the same problem that I see when you work on the fits file from the link?

I've attached my StarTools.log file.

cytan

[admin]

And here's the checksum of the resources file and size:
Is this what you'd expect? I double checked that the resources file has the same checksums as that in the zip distribution.

That all looks in order!

Do you see the same problem that I see when you work on the fits file from the link?

I can no longer replicate this and that's the strange thing... The GPU load is now so granularized that processing the load should not be a problem any more.

Since your system is also a dual graphics solution, you could try switching between iGPU for display duties and freeing up the 5500M for compute (see here). (or the other way around of course)

Let me know if that makes any difference!

[cytan299]

And here's the checksum of the resources file and size:
Is this what you'd expect? I double checked that the resources file has the same checksums as that in the zip distribution.

That all looks in order!

Do you see the same problem that I see when you work on the fits file from the link?

I can no longer replicate this and that's the strange thing... The GPU load is now so granularized that processing the load should not be a problem any more.

Since your system is also a dual graphics solution, you could try switching between iGPU for display duties and freeing up the 5500M for compute (see here). (or the other way around of course)

Let me know if that makes any difference!

Hi Ivo,
I just tried both "0" and "1" in the opencldeviceindex.cfg file. Is there some indication that the 5500M has been freed up for computation? I looked at the About panel and there's nothing obvious.

cytan

[admin]

Hi Ivo,
I just tried both "0" and "1" in the opencldeviceindex.cfg file. Is there some indication that the 5500M has been freed up for computation? I looked at the About panel and there's nothing obvious.

I just found this application for macOS;

https://codyschrank.github.io/gSwitch/

explained here

https://www.addictivetips.com/mac-os/sw ... gpu-macos/

This may hold some promise for your system.

The idea is to switch macOS to use Integrated Only (for display duties), thereby freeing up the discrete GPU (for compute).

The screenshots at addictivetips also mention a Recovery_Timeouts feature. This may be a way to change the watchdog timeout? (though I'm still not 100% sure this is the issue right now in your particular case)

[cytan299]

Hi Ivo,
I just tried both "0" and "1" in the opencldeviceindex.cfg file. Is there some indication that the 5500M has been freed up for computation? I looked at the About panel and there's nothing obvious.

I just found this application for macOS;

https://codyschrank.github.io/gSwitch/

explained here

https://www.addictivetips.com/mac-os/sw ... gpu-macos/

This may hold some promise for your system.

The idea is to switch macOS to use Integrated Only (for display duties), thereby freeing up the discrete GPU (for compute).

The screenshots at addictivetips also mention a Recovery_Timeouts feature. This may be a way to change the watchdog timeout? (though I'm still not 100% sure this is the issue right now in your particular case)

I'll install and try it out. But there's one other possibility, if I understand from the FAQ in the first link:

*************
Why does the app go back to dynamic switching when a display is plugged in?

Unfortunately your mac is designed such that in order to use an external display, it has to use the discrete graphics card. And since you plugged in the cable I’m assuming you want to use the display.
**************

It seems that the GPU will necessarily be used when I use an external display. I *was* using an external display when I was ST. So, I think the easiest thing to do first is to use my MacBook Pro display and not my external monitor first to see if the problem goes away.

cytan