From owner-freebsd-current Tue Jun 19 12:46:22 2001 Delivered-To: freebsd-current@freebsd.org Received: from gull.mail.pas.earthlink.net (gull.mail.pas.earthlink.net [207.217.121.85]) by hub.freebsd.org (Postfix) with ESMTP id 7809F37B401 for ; Tue, 19 Jun 2001 12:46:20 -0700 (PDT) (envelope-from tlambert2@mindspring.com) Received: from mindspring.com (dialup-209.244.104.1.Dial1.SanJose1.Level3.net [209.244.104.1]) by gull.mail.pas.earthlink.net (EL-8_9_3_3/8.9.3) with ESMTP id MAA22016; Tue, 19 Jun 2001 12:44:06 -0700 (PDT) Message-ID: <3B2FABA8.D439B306@mindspring.com> Date: Tue, 19 Jun 2001 12:44:40 -0700 From: Terry Lambert Reply-To: tlambert2@mindspring.com X-Mailer: Mozilla 4.7 [en]C-CCK-MCD {Sony} (Win98; U) X-Accept-Language: en MIME-Version: 1.0 To: Giorgos Keramidas Cc: Kris Kennaway , Peter Wemm , Steve Kargl , current@FreeBSD.ORG Subject: Re: convert libgmp to a port? References: <20010616201135.A60010@troutmask.apl.washington.edu> <20010617063845.E53DA39E9@overcee.netplex.com.au> <20010617182256.A17724@hades.hell.gr> <20010617135156.B94632@xor.obsecurity.org> <20010618010338.A50655@hades.hell.gr> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-freebsd-current@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG Giorgos Keramidas wrote: > > On Sun, Jun 17, 2001 at 01:51:56PM -0700, Kris Kennaway wrote: > > > libbn is already part of OpenSSH; it's a trivial matter to make it > > into a standalone library. In other words, we already include two > > functionally equivalent bignum libraries in FreeBSD, so one of them > > should go. > > I couldn't agree more :) I'm going to word this strongly, mostly because I feel strongly about the underlying issues. The SSL one is known to be very slow, and was written as a proof of concept by the author. Please read the release notes; it is seriously slow. Replacing it will increase your SSL performance significantly. OpenSSL is similarly very slow, and not very well architected; the use of pointers in the hardware acceleration code just about doubles the number of instructions required to prepare an argument descriptor. There is much low hanging fruit in that code. It is mostly bad code, and its reliance on threading instead of finite state automata shows a serious lack of understanding of what it takes to make high performance code -- unfortunately, it is becoming a more common trade off, since it is much easier to write threads than it is to seperate statites, and keep the code understandable to a college sofphmore contributor base: it is becoming a disease in Open Source software in general. Please _do not_ "standardize" on "equivalents" without benchmarks -- the functionality may be equivalent, but the performance certainly is not. -- Terry To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message