On Tuesday 13 December 2011, William Stein wrote: > Thoughts? Are there any experts out there in code benchmarking done > from a more mathematically sophisticated perspective than just one > number? It's entirely possible I'm making some stupid mistake in > suggesting the above.
Carlo Wood implemented pretty nice benchmarketing code for us in M4RI https://bitbucket.org/malb/m4ri/src/ec6d607ba523/testsuite/benchmarking.c * Example usage: * * ./bench_elimination -s 0 -m 4 -c 90 -a 0.005 -d -t 30 -n 1000 1000 1000 * * would run at most 30 seconds (-t) or 1000 times (-n), whichever comes * first, or stop after the real average of wall time (-s 0) falls with 90% * certainty (-c) in a range that is +/- 0.005 times the observed mean (-a: * accuracry), but no sooner than that at least 4 (-m: minimum) measurements * have been done. It would also print (-d: dump) each measurement * (0:microseconds * 1:cpuclocks). It uses the "Student-t" method to estimate the certainty (don't ask me what that is, I don't know). Cheers, Martin -- name: Martin Albrecht _pgp: http://pgp.mit.edu:11371/pks/lookup?op=get&search=0x8EF0DC99 _otr: 47F43D1A 5D68C36F 468BAEBA 640E8856 D7951CCF _www: http://martinralbrecht.wordpress.com/ _jab: martinralbre...@jabber.ccc.de -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
