mmfausnaugh / tica

TICA: TESS Image CAlibrator
MIT License
6 stars 0 forks source link

Pythonic Parallelization of WCS #13

Closed mmfausnaugh closed 1 year ago

mmfausnaugh commented 2 years ago

We'll need to change step1_get_refimg_ctrlpts.py and step2_mkwcs.py to use Pool, so that we can throw arbitrary numbers of cores at this in EM2 (which as 200 sec FFIs --> 3x as miuch data).

mmfausnaugh commented 2 years ago

Can probably also improve step1 speed by making these local querries of tic 8.2

mmfausnaugh commented 1 year ago

fixed in version 1.3.0. Interesting, not as much speed as one might think. Experimentation found that .apply_async could get about 1000 seconds faster. Can get individual CCDs down to 200sec using 60 cores, but needing to do 16 of these is just as fast as doing 16 python processes with 4 CCDs each.

This is pushing into issues with our harddware and IO system, so keeping the flexible framework and marking as done.