Date: Mon, 16 Dec 1996 00:58:19 -0800 (PST) From: David Muir Sharnoff <muir@idiom.com> To: Heikki Suonsivu <hsu@clinet.fi> Cc: freebsd-bugs@freebsd.org Subject: Re: Re: Re: bin/146 Message-ID: <199612160858.AAA02647@idiom.com>
next in thread | raw e-mail | index | archive | help
* gzip is by default doing best compression ratio, not best speed. Compress * performance and memory use can be tuned also, using -b option, but * compression performance drops dramatically while gzip options do not make * that much a difference (gzip -1 still compresses better than compress at * its best, and result is half of what compress -b12 makes!): * * time ratio decompress * ncompress 4.2.4 -b12 6.80+1.06 60.4% 3.68+1.27 * gzip -1 10.49+0.68 31.4% 3.46+0.21 * ncompress 4.2.4 13.01+1.48 46.9% 4.12+1.11 * /usr/bin/compress 18.45+0.89 43.6% 6.68+0.31 * gzip 27.98+0.65 26.8% 3.25+0.18 Nice table! * For people who need the best compression speed I would rather suggest gzip * -1. I cannot really see why anyone would want to use slightly faster * compress when compression result is that much worse ? The only time it becomes important is when you are trying to stream a tape. In times past, I've been able to stream tapes using ncompress 4.2.4 but not gzip. In my current setup, my systems are fast enough to stream tapes with gzip -1. As for you tests, I find that that compression ration varies quite a bit depending on which partition I'm dumping. Also speed. Gzip -1 can be faster than ncompress (as your test shows) but the reverse also can happen. The only reason I brought this up is that if FreeBSD has a compress program at all, it might as well have the best... -Dave
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199612160858.AAA02647>