Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!mnetor!seismo!mcvax!unido!iaoobelix!wagner From: wagner@iaoobelix.UUCP Newsgroups: comp.sources.d Subject: Fastest `grep'?? Really? - (nf) Message-ID: <7200003@iaoobelix.UUCP> Date: Sun, 21-Jun-87 11:35:00 EDT Article-I.D.: iaoobeli.7200003 Posted: Sun Jun 21 11:35:00 1987 Date-Received: Fri, 26-Jun-87 03:35:51 EDT Lines: 33 Nf-ID: #N:iaoobelix:7200003:000:1457 Nf-From: iaoobelix!wagner Jun 21 16:35:00 1987 A few weeks ago, the "fastest grep around" was posted thru comp.sources. Well, reading this announcement I was really interested in the preformance figures of this nice program. So I just moved it over to a Sun 3/260 (SunOS 3.2) and compiled it as it came with 'notes'. Ahemm! Well! **SURPRISE** The results I obtained from just little playing around with this reportedly fastest grep program were a bit :-) :-) disappointing. Below you find the figures I got from doing a time grep "[0-9][a-z] [a-z]" /usr/man/cat2/* > /dev/null with several grep versions. I say several versions (not just two) because there are three of them: (a) the standard /bin/grep coming with UNIX, (b) the grep from comp.sources, and (c) the grep from an older posting of Ozan S. Yigit (utzoo!yetti!oz) coming with the author's own version of the regexp package (here, it is called re_exp). Ok, here are the timing figures: (a) time: 6.7u + 0.9s = 60% of 0:12 (b) time: 29.7u + 1.3s = 89% of 0:34 (c) time: 6.5u + 1.0s = 70% of 0:10 Remember: It is a Sun 3/260, and I am the only user; nobody's distracting my machine from grepping thru /usr/man/cat2... Is there anybody out there who knows why some people called the program (a) the "fastest grep around"?? Apparently, it is the factor four (in average) slower even than the standard `grep'... Juergen Wagner, (USENET) ...seismo!unido!iaoobel!wagner ("Gandalf") Fraunhofer Institute IAO, Stut7=EENV