From owner-freebsd-current Thu Aug 17 23:33:54 1995 Return-Path: current-owner Received: (from majordom@localhost) by freefall.FreeBSD.org (8.6.11/8.6.6) id XAA04058 for current-outgoing; Thu, 17 Aug 1995 23:33:54 -0700 Received: from haywire.DIALix.COM (haywire.DIALix.COM [192.203.228.65]) by freefall.FreeBSD.org (8.6.11/8.6.6) with ESMTP id XAA04004 for ; Thu, 17 Aug 1995 23:33:30 -0700 Received: (from news@localhost) by haywire.DIALix.COM (8.7.Beta.11/8.7.Beta.11/DIALix) id OAA11524 for freebsd-current@freebsd.org; Fri, 18 Aug 1995 14:33:14 +0800 (WST) Received: from GATEWAY by haywire.DIALix.COM with netnews for freebsd-current@freebsd.org (problems to: usenet@haywire.dialix.com) To: freebsd-current@freebsd.org Date: 18 Aug 1995 14:33:10 +0800 From: peter@haywire.dialix.com (Peter Wemm) Message-ID: <411c76$b81$1@haywire.DIALix.COM> Organization: DIALix Services, Perth, Australia. References: <199508171335.GAA20286@time.cdrom.com>, <199508172048.WAA10784@uriah.heep.sax.de> Subject: Re: pppd Sender: current-owner@freebsd.org Precedence: bulk j@uriah.heep.sax.de (J Wunsch) writes: >As Jordan K. Hubbard wrote: >> >> The biggest difference I can see is that 2.2 supports some sort of >> compression called "BSD-compress", whatever the heck that is.. >Ain't this the ``predictor 1'' compression? Would be nice to have it. No. It's a compress(1) modified to run as a data stream. Considering that v42bis is a similar routine, and most modems only do 11 bit compression, and this code can do up to 15 bits (if you are prepared to allocate 64K of kernel memory for the tables), it has been demonstrated to cause significant performance boosts (and, suprisingly, latency drops.) A 14.4K modem with compression off will typically do something in the order of 180ms ping times. When youy enable v42bis, that typically goes up by 40ms. I believe I have saved a copy of an article from the ppp mailing list where the rtt went *down* when kernel compression was enabled, and the actual cpu overhead was reduced, with the cost of compression being less than the cost of the serial IO that was reduced. -Peter >-- >cheers, J"org >joerg_wunsch@uriah.heep.sax.de -- http://www.sax.de/~joerg/ >Never trust an operating system you don't have sources for. ;-)