From mboxrd@z Thu Jan 1 00:00:00 1970 Received: (from majordomo@localhost) by pauillac.inria.fr (8.7.6/8.7.3) id VAA11869; Thu, 10 Oct 2002 21:46:00 +0200 (MET DST) X-Authentication-Warning: pauillac.inria.fr: majordomo set sender to owner-caml-list@pauillac.inria.fr using -f Received: from concorde.inria.fr (concorde.inria.fr [192.93.2.39]) by pauillac.inria.fr (8.7.6/8.7.3) with ESMTP id VAA11875 for ; Thu, 10 Oct 2002 21:46:00 +0200 (MET DST) Received: from snipe.mail.pas.earthlink.net (snipe.mail.pas.earthlink.net [207.217.120.62]) by concorde.inria.fr (8.11.1/8.11.1) with ESMTP id g9AJjw505562 for ; Thu, 10 Oct 2002 21:45:59 +0200 (MET DST) Received: from user-0ccegbj.cable.mindspring.com ([24.199.65.115] helo=dragonfly.localdomain) by snipe.mail.pas.earthlink.net with smtp (Exim 3.33 #1) id 17zjFt-0004Mh-00 for caml-list@inria.fr; Thu, 10 Oct 2002 12:45:57 -0700 Received: from 209-9-234-140.sdsl.cais.net ([209.9.234.140]) (SquirrelMail authenticated user yminsky) by dragonfly.localdomain with HTTP; Thu, 10 Oct 2002 15:45:57 -0400 (EDT) Message-ID: <40416.209.9.234.140.1034279157.squirrel@dragonfly.localdomain> Date: Thu, 10 Oct 2002 15:45:57 -0400 (EDT) Subject: Re: [Caml-list] Num library From: "Yaron M. Minsky" To: In-Reply-To: <200210101714.TAA07335@pauillac.inria.fr> References: <23100.209.9.234.140.1034261810.squirrel@dragonfly.localdomain> from "Yaron M. Minsky" at "Oct 10, 102 10:56:50 am" <200210101714.TAA07335@pauillac.inria.fr> X-Priority: 3 Importance: Normal X-MSMail-Priority: Normal X-Mailer: SquirrelMail (version 1.2.6) MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit Sender: owner-caml-list@pauillac.inria.fr Precedence: bulk I thought I was mostly passing on the common wisdom about the Num library. I didn't test it myself, but the numerix documentation has some performance numbers which make the Big_int part of Num look quite weak compared to GMP and Numerix --- it came up as the loser over both in most cases, especially for large sizes. You can find it on Michel Quercia's page. But I don't know much about Num's performance outside of Michel's results, and I'm open to being corrected. y >> Num is by all accounts a pretty mediocre bignum implementation. >> There's also mlgmp, which is an interface to GMP. I haven't used it, >> but GMP is fast. There's also Numerix, which I have used. Numerix at >> least at one point was a good deal faster than GMP in many cases, and >> is very easy to use. I'm not sure how GMP and Numerix compare in >> terms of speed these days, since GMP has seen more development and >> Numerix has not. >> y > > I'm afraid this is a bit rough, not a fair scientific study. > > Once more, just one question to you: which layer of the Num library had > you benchmarked ? > > Pierre Weis > > INRIA, Projet Cristal, Pierre.Weis@inria.fr, > http://pauillac.inria.fr/~weis/ -- |--------/ Yaron M. Minsky \--------| |--------\ http://www.cs.cornell.edu/home/yminsky/ /--------| Open PGP --- KeyID B1FFD916 (new key as of Dec 4th) Fingerprint: 5BF6 83E1 0CE3 1043 95D8 F8D5 9F12 B3A9 B1FF D916 ------------------- To unsubscribe, mail caml-list-request@inria.fr Archives: http://caml.inria.fr Bug reports: http://caml.inria.fr/bin/caml-bugs FAQ: http://caml.inria.fr/FAQ/ Beginner's list: http://groups.yahoo.com/group/ocaml_beginners