search

UKZN-Astronomy / corrcal

Python/C code for calibration of quasi-redundant arrays. Different (fixed) algorithm from original corrcal
BSD 2-Clause "Simplified" License
1 stars 0 forks source link

Comparing Cholesky and BLAS/LAPACK decomposition #14

Closed piyanatk closed 4 years ago

piyanatk commented 5 years ago

Originally posted by @ronniyjoseph in #5 (comment)

@sievers' arXiv pre-print discusses how c-code actually slows corrcal down for small problems:

For small (∼ 100 antenna) problems with only a few source/redundant vectors, using BLAS/LAPACK actually slows down the code, since the function overhead for e.g. Cholesky decomposition of 2x2 blocks is non-negligible. So, the reference implementation fully self-contained.

@Mdlalose, @Zahra-Kader, @ronniyjoseph Have you run tests to confirm this? If not, we should put it in a to do list.

ronniyjoseph commented 5 years ago

@piyanatk I have not tested this myself, I've only run the code on the 144 antenna example supplied in the example. And a 36 hexagon, but I did not do speed comparisons

piyanatk commented 4 years ago

According to Jon, BLAS is not being used in the current C codes.

Mdlalose commented 4 years ago

I'll put it on do list.

On Tue, Dec 10, 2019, 18:52 Piyanat (Boom) Kittiwisit < notifications@github.com> wrote:

According to Jon, BLAS is not being used in the current C codes.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/UKZN-Astronomy/corrcal/issues/14?email_source=notifications&email_token=ADDW4VL64JMUISGZCQYBYOTQX7CNVA5CNFSM4JAYTFYKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEGP6G5A#issuecomment-564126580, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADDW4VLWCXZUFXPF6C524HTQX7CNVANCNFSM4JAYTFYA .

piyanatk commented 4 years ago

Dropping this issue since blas has not been implemented in the current code.