Re: Compiler optimizations and other performance hacks for HiGHS
Søren Fuglede Jørgensen via SciPy-User <[email protected]> Tue, 14 Mar 2023 15:18:21 -0000
| Newsgroups | gmane.comp.python.scientific.user |
|---|---|
| Message-ID | <[email protected]> |
Thanks Ralf, that's a great help! Given your hint, I tried running `python dev.py ipython` once while sticking to `'buildtype=debugoptimized'` and once after cleaning the repository and changing to `'buildtype=release'` in line 11 of the top-level `meson.build`. In both cases, I ran `linprog` on a large-ish input (A_eq shape = (7890, 59761), nnz = 59761, A_ub shape = (358, 59761), nnz = 541705) and found no improvement in performance between the two cases; if anything, `debugoptimized` was slightly faster. So assuming I got it right, that particular trick didn't turn out to be useful in my case. Good to know regardless! _______________________________________________ SciPy-User mailing list -- [email protected] To unsubscribe send an email to [email protected] https://mail.python.org/mailman3/lists/scipy-user.python.org/ Member address: [email protected]