New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Get gf2x version 1.1 into Sage! #2114
Comments
comment:1
The "attached version" is here: |
comment:2
A new version (0.3.1) is available from http://wwwmaths.anu.edu.au/~brent/gf2x.html. |
comment:4
GF2X has now its own development page: http://gf2x.gforge.inria.fr/. |
comment:5
Note that since ntl-5.5, NTL can now be configured to use GF2X instead of its own routines. |
comment:7
Paul, I'm finally packaging this, only planning to build NTL on top of it, no "native" interface. |
Changed keywords from none to spkg gf2x |
Author: Jean-Pierre Flori |
This comment has been minimized.
This comment has been minimized.
comment:8
Upped spkgs, with the GF2X_TUNE option off by default (took one hour on a quite recent Xeon (with only one thread, but I did not test whether tuning gets parallelized)). I did not check this actually speeds up NTL, anyone wanting to benchmark the new NTL spkg against the old one please go ahead. As NTL is a standard spkg, not sure what the way to go is. |
Spkg diff, for review only. |
comment:9
Attachment: ntl-5.5.2.p1.diff.gz Replying to @jpflori:
Why should I? You should decide if you want this: |
comment:10
I say we want this as a standard package and would prefer to avoid an optional stage (just imagine we patch ntl inbetween. |
comment:11
So does it actually speed up things, considering that we disable tuning? If not then forget about it. If yes then I'd be happy to see it included. |
comment:12
Just for hint, my experience playing with FLINT, using gf2x for GF(2) polynomials instead of the basic implementation inside FLINT using long gave an incredible speedup. |
comment:13
Replying to @jpflori:
Jean-Pierre, make tune-lowlevel is still highly recommended, since it will choose the best routine up to 9 words. Then you can do say Paul |
Work Issues: tuning |
comment:14
Ok, I'll do that. |
comment:15
Replying to @jpflori:
On my pc, multiplying two random elements of GF(2**10000) goes from 103us in Sage 5.9 with old NTL to 21.4us in 5.10.rc0 with NTL+gf2x. |
comment:54
Replying to @nexttime:
Leif, thanks, I've fixed a few upstream. It is always nice to get feedback when a package is used in Sage! Paul |
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
comment:57
I made a few small changes to |
comment:58
For NTL, I added the |
This comment has been minimized.
This comment has been minimized.
Reviewer: Jeroen Demeyer |
comment:61
Replying to @jdemeyer:
You didn't fix "gf2x is a C/C+ software package" though... ;-) Could you attach a diff of your changes? I don't want to re-review everything. Retesting is odd enough. |
Spkg diff, for review only. |
Attachment: gf2x-1.1.diff.gz |
comment:62
Attachment: gf2x-1.1-jdemeyer.diff.gz Replying to @nexttime:
Fixed now.
|
comment:63
Replying to @jdemeyer:
I would have dropped the "C++"; if gf2x was (partially) implemented in C++, we'd have to set / change I'd probably also use |
comment:64
Did I mention it also builds with |
comment:67
the C++ part of gf2x is only in the "apps" subdirectory, which contains binaries to be linked with NTL Paul |
comment:68
Replying to @zimmermann6:
We can't build those apps (at least not immediately), since we decided to let NTL depend on gf2x. |
comment:69
Positive review to everything except my changes (attachment: gf2x-1.1-jdemeyer.diff). |
comment:70
I'm ok with them. |
comment:71
FWIW, I was going to give the previous spkgs positive review right when Jeroen started to change both, so I don't insist on fixing the hardcoded |
Merged: sage-5.11.beta1 |
Check out http://wwwmaths.anu.edu.au/~brent/gf2x.html
It's:
Latest 1.1 version is at http://gf2x.gforge.inria.fr/
Use spkgs at:
Apply to Sage's root:
Depends on #12447
CC: @zimmermann6 @malb @jpflori @jdemeyer
Component: packages: standard
Keywords: spkg gf2x
Author: Jean-Pierre Flori
Reviewer: Jeroen Demeyer
Merged: sage-5.11.beta1
Issue created by migration from https://trac.sagemath.org/ticket/2114
The text was updated successfully, but these errors were encountered: