Re: [fricas-devel] FiniteField(Extension)-Problem
On 11/19/19 5:49 AM, Ralf Hemmecke wrote: >> To answer question "what/why takes time" use profiler. sbcl >> profiler shows that time is spent in factorizer. > > Thank you. I think that helps me to continue. > > Still I would be interested in how exactly I would have use the profiler > and how to interpret the output. > > I used what Qian suggested some time ago > > )lisp (require :sb-sprof) > )lisp (sb-sprof:start-profiling) > -- fricas computation here > coerce(c2)$F4 > )lisp (sb-sprof:stop-profiling) > )lisp (sb-sprof:report) > > But couldn't exactly make sense out of the output. > As far as I can tell it just counts the number of times a certain > function is called, but not the time it spents in a function. > > Ralf > The "Count" you see is not the number of times a certain function is called, instead, the calling stack is interrupted and examined periodically to see which function is being called: so this is a statistical method to get how much time each function is spending -- the time spent by a function is proportional to the count it gets during those examination. Also notice the difference between "Count" and "Total Count", "Count" means how much time is spent by function body, "Total Count" means how much time is spent by function body and sub-function calls. -- You received this message because you are subscribed to the Google Groups "FriCAS - computer algebra system" group. To unsubscribe from this group and stop receiving emails from it, send an email to fricas-devel+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/fricas-devel/88f219ba-2dac-4821-d14f-106eca991ebf%40gmail.com.
Re: [fricas-devel] FiniteField(Extension)-Problem
> To answer question "what/why takes time" use profiler. sbcl > profiler shows that time is spent in factorizer. Thank you. I think that helps me to continue. Still I would be interested in how exactly I would have use the profiler and how to interpret the output. I used what Qian suggested some time ago )lisp (require :sb-sprof) )lisp (sb-sprof:start-profiling) -- fricas computation here coerce(c2)$F4 )lisp (sb-sprof:stop-profiling) )lisp (sb-sprof:report) But couldn't exactly make sense out of the output. As far as I can tell it just counts the number of times a certain function is called, but not the time it spents in a function. Ralf -- You received this message because you are subscribed to the Google Groups "FriCAS - computer algebra system" group. To unsubscribe from this group and stop receiving emails from it, send an email to fricas-devel+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/fricas-devel/f6784196-c23b-1bdb-cc5e-c17ab84dbfbd%40hemmecke.org.
Re: [fricas-devel] AXIOM --> FRICAS
> There is still weirdness in src/doc/htex/ug15.htex, in section > > Old News about AXIOM Version 2.x > > changing Axiom to FriCAS for me looks like reading 1984 (for > example "correcting" past issues of newspapers to conform to > changed party policy). Oh. I liked that part of "1984" very much. It makes clear that history is not as stable as we think. Clearly, I don't want to be part of that. I now mostly left that part under "Old New" as it was. I only changed \axiomxl to AXIOM-XL in two places, since I've redefined \axiomxl to produce Aldor in the .sty file. > I would keep this part of ug15.htex as is. If you agree > commit the rest. Yep. Thanks. Ralf -- You received this message because you are subscribed to the Google Groups "FriCAS - computer algebra system" group. To unsubscribe from this group and stop receiving emails from it, send an email to fricas-devel+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/fricas-devel/16d2e30e-87c9-1adb-a99d-8d1cb2721891%40hemmecke.org.
Re: [fricas-devel] FiniteField(Extension)-Problem
Ralf Hemmecke wrote: > > Hello Johannes, > > as one of the authors of the finite field implementation you might > probably be able to tell me what is happening here, see attachment. > > The last > > coerce(c2)$F4 > > takes nearly 18 seconds on my laptop (first time only, of course). > That's probably due to the computation of the discrete logarithm table. To answer question "what/why takes time" use profiler. sbcl profiler shows that time is spent in factorizer. Namely to really build F4 we need an irreducible polymomial. We compute it by factoring candidates. It is possible that smarter selection of candidates would limit work in factoring. But the main point is that ATM factoring over finite field extentions is slow (over prime fields we use much better routine). -- Waldek Hebisch -- You received this message because you are subscribed to the Google Groups "FriCAS - computer algebra system" group. To unsubscribe from this group and stop receiving emails from it, send an email to fricas-devel+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/fricas-devel/E1iWjKS-0007sZ-R0%40hera.math.uni.wroc.pl.
Re: [fricas-devel] AXIOM --> FRICAS
Ralf Hemmecke wrote: > > I've amended the axiom branch. No changes to any of the Makenotes.tex > anymore. > > Otherwise the difference between my previous commit and the one I want > to eventually commit to SVN is attached. > > Can I commit > > https://github.com/hemmecke/fricas/commit/b71a9d46170488eb5e524894ed3f379dcac25f6e > > to SVN? > There is still weirdness in src/doc/htex/ug15.htex, in section Old News about AXIOM Version 2.x changing Axiom to FriCAS for me looks like reading 1984 (for example "correcting" past issues of newspapers to conform to changed party policy). When changing names I kept original name as this text was taken verbatim. In general in "news" files corrections should be minimal only for obvoius errors (preferably none). I would keep this part of ug15.htex as is. If you agree commit the rest. -- Waldek Hebisch -- You received this message because you are subscribed to the Google Groups "FriCAS - computer algebra system" group. To unsubscribe from this group and stop receiving emails from it, send an email to fricas-devel+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/fricas-devel/E1iWiuU-0007jd-Ai%40hera.math.uni.wroc.pl.