[sage-devel] Re: Proposal for faster linear algebra over GF(q), q<255 odd and non-prime

2011-12-01 Thread Dima Pasechnik
On Wednesday, November 30, 2011 10:52:14 PM UTC+8, Simon King wrote: > > Hi Dima, > > On 30 Nov., 15:29, Dima Pasechnik wrote: > > I might get blamed for making discouraging remarks, but let me play the > > devil's advocate: > > > > I wonder if these kinds of speed-ups are to be beaten, soon, by

[sage-devel] Re: Proposal for faster linear algebra over GF(q), q<255 odd and non-prime

2011-12-01 Thread Simon King
Hi Tom, On 1 Dez., 10:08, Tom Boothby wrote: > Yeah, Sage has one implemented in Cython.  Use MeatAXE for the base > case of a reasonable size based on the dimension / prime. I know about that. Actually, I wrote the current generic implementation of Strassen-Winograd multiplication in Sage, foll

Re:[sage-devel] Re: Proposal for faster linear algebra over GF(q), q<255 odd and non-prime

2011-12-01 Thread Martin Albrecht
Hi, LinBox uses the BLAS for dense matrices mod p where p <= 2^23, hence it would benefit from advances in that direction. There is also some (experimental?) code for dealing with non-prime fields in LinBox. Furthermore, there is some code - AFAIK not in LinBox yet - for packing multiple primes int

Re: [sage-devel] Re: Proposal for faster linear algebra over GF(q), q<255 odd and non-prime

2011-12-01 Thread Martin Albrecht
Tom, Simon's MeatAxe fork implements its own Strassen. But perhaps Sage's asymptotically fast Gaussian elimination could be useful? On Dec 1, 2011 9:09 AM, "Tom Boothby" wrote: > On Wed, Nov 30, 2011 at 6:52 AM, Simon King > wrote: > > Hi Dima, > > > > On 30 Nov., 15:29, Dima Pasechnik wrote: >

Re: [sage-devel] Re: Proposal for faster linear algebra over GF(q), q<255 odd and non-prime

2011-12-01 Thread Tom Boothby
On Wed, Nov 30, 2011 at 6:52 AM, Simon King wrote: > Hi Dima, > > On 30 Nov., 15:29, Dima Pasechnik wrote: >> I might get blamed for making discouraging remarks, but let me play the >> devil's advocate: >> >> I wonder if these kinds of speed-ups are to be beaten, soon, by >> sufficiently fast har

[sage-devel] Re: Proposal for faster linear algebra over GF(q), q<255 odd and non-prime

2011-11-30 Thread Simon King
Hi Dima, On 30 Nov., 15:29, Dima Pasechnik wrote: > I might get blamed for making discouraging remarks, but let me play the > devil's advocate: > > I wonder if these kinds of speed-ups are to be beaten, soon, by > sufficiently fast hardware implementations of level 2 and 3 BLAS, coupled > with so

[sage-devel] Re: Proposal for faster linear algebra over GF(q), q<255 odd and non-prime

2011-11-30 Thread Dima Pasechnik
I might get blamed for making discouraging remarks, but let me play the devil's advocate: I wonder if these kinds of speed-ups are to be beaten, soon, by sufficiently fast hardware implementations of level 2 and 3 BLAS, coupled with some crude use of their super-fast vector arithmetic. Assumin

[sage-devel] Re: Proposal for faster linear algebra over GF(q), q<255 odd and non-prime

2011-11-30 Thread Simon King
Hi Jeroen, On 30 Nov., 14:48, Jeroen Demeyer wrote: > The easiest solution here would be making meataxe a standard spkg > directly.  Ask William whether you can do this... I guess making it standard would require to (1) make it an optional package first, and (2) have a vote on sage-devel. Cheer