Slow to platesolve

[Stefan Jagger]

Any ideas why platesolving takes 6 minutes please? The CPU and Mem usage on the machine are low. I've placed a space either side of the command which seems to take a while to complete. I have all the data from ( http://broiler.astrometry.net/~dstn/4200/ ) so I was thinking maybe it's too much data to solve from?

Many thanks,

Stefan

2021-05-24 16:54:33,603 Starting Job processing for Job 1

2021-05-24 16:54:33,669 Creating directory /astrometry.net/net/data/jobs/0000/00000001

2021-05-24 16:54:33,671 submission id 1

2021-05-24 16:54:33,676 running: augment-xylist --downsample 2 --tweak-order 2 --scale-units degwidth --scale-low 0.1 --wcs wcs.fits --out /astrometry.net/net/data/jobs/0000/00000001/job.axy --image /astrometry.net/net/data/files/uploaded/cae/caea90b6c8323c9963aeb467b3347f264682aa64 --rdls rdls.fits --scale-high 180.0 --corr corr.fits 

2021-05-24 16:54:50,670 created axy file /astrometry.net/net/data/jobs/0000/00000001/job.axy

2021-05-24 16:54:50,670 command: cd /astrometry.net/net/data/jobs/0000/00000001 && /astrometry.net/net/solve_script.sh job-test-1 job.axy >> /astrometry.net/net/data/jobs/0000/00000001/log

2021-05-24 17:00:25,405 Solver completed successfully.

2021-05-24 17:00:25,405 Checking for WCS file /astrometry.net/net/data/jobs/0000/00000001/wcs.fits

2021-05-24 17:00:25,406 WCS file exists

2021-05-24 17:00:25,456 Created TanWCS: <TanWCS: CRVAL (151.469136, 68.733816) CRPIX (5696.234741, 3841.165710) CD (0.000315, 0.000431; -0.000431 0.000315) Image size (8288.000000, 5644.000000)>

2021-05-24 17:00:25,494 SkyLocation: <SkyLocation: nside(16) healpix(475)>

2021-05-24 17:00:25,521 Created Calibration Calibration 1

2021-05-24 17:00:27,858 Finished job 1

[Dustin Lang]

The index files get read from disk as required and then cached, so if you have more memory in your computer than the size of the index files being used, then once they all get read in, it should run fast.  If you re-run a field, is it faster?

If you have more index files than memory, then life is more difficult.

cheers,

--dustin

--
You received this message because you are subscribed to the Google Groups "astrometry" group.
To unsubscribe from this group and stop receiving emails from it, send an email to astrometry+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/astrometry/7695acfc-e36f-49f0-8ecd-0eb30b2aa107n%40googlegroups.com.

[Stefan Jagger]

Hi Dustin,

Thanks for your reply.  That's a lot of memory when using all the index files. Yes it was a bit quicker on the re-run and quicker on a jpg rather than fits file too I suspect cause the of data quality.

Is it possible to use a fraction of the index files? I was thinking of using just index-4202-** ?  I'm trying to understand http://astrometry.net/doc/readme.html#index-building-programs but am struggling to work out the right data i'd need for approx. 50-300mm focal lengths.

Cheers,

Stefan

[Stefan Jagger]

Looks like my docker instance is only using 1GB of the total 4GB available . Can you provide any pointers to where I might look to increase the RAM limit?

On Monday, 24 May 2021 at 22:47:52 UTC+1 dstn...@gmail.com wrote:

[Bryan]

Stefan

You can limit the index files used in the solve-field command-line with scale-low and scale-high.  You can do the same in the /usr/local/astrometry/etc/astrometry.cg/  file.  However, if you change cameras/lenses/scopes a lot, then the solve-field approach is more flexible; although you can use the back-end config file (astrometry.cfg) for multiple equipment setups.

Calculate the field of view (FOV) for your setup, then set --scale-low about 1/2 that and --scale--high about twice that.  You can change the units of FOV with the --

scale-units parameter in solve field.

See Tricks and Tips in https://astrometry.net/doc/readme.html#index-building-programs

Bryan

[Dustin Lang]

Also, if you know roughly where you are on the sky, --ra, --dec, and --radius will allow loading only that broad chunk of sky (for the files that are of the form index-42xx-yy.fits, it'll only load the yy's required).

To view this discussion on the web visit https://groups.google.com/d/msgid/astrometry/e59b4af2-4be8-4f5b-aae4-3582a890b2ebn%40googlegroups.com.

[Stefan Jagger]

Thanks guys I appreciate your help.