Multicore

11 views
Skip to first unread message

Martin R. Albrecht

unread,
May 27, 2020, 9:10:18 AM5/27/20
to Fplll-Devel, Artur Miguel
Hi all,

CC: Artur, who recently published some (multicore) comparisons between FPLLL and his LUSA library: https://eprint.iacr.org/2020/605

I have solved the mystery of why we didn’t see speed ups when enabling multiple cores:

https://github.com/fplll/fpylll/commit/deb9c85bcd1b6e39a2512e74578d3d685d107394

With this advice applied, I’m seeing:

1 Thread:

__main__: = block size: 80, m: 32 =
__main__: t: 114.3077s, w: 114.3109s, Strategy< 80, (58), 0.08-0.25>, 1.01020

2 Threads:

__main__: = block size: 80, m: 32 =
__main__: t: 110.2549s, w: 61.8496s, Strategy< 80, (58), 0.08-0.25>, 1.01021

4 Threads:

__main__: = block size: 80, m: 32 =
__main__: t: 113.2351s, w: 39.1968s, Strategy< 80, (58), 0.08-0.25>, 1.01022

Above, “t” is cputime, “w” is walltime. I’ll play around a bit to see how much computing strategies targeting multiple cores gets.

Cheers,
Martin


--

_pgp: https://keybase.io/martinralbrecht
_www: https://malb.io

Reply all
Reply to author
Forward
0 new messages