Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.2 9/18/84; site brl-tgr.ARPA Path: utzoo!watmath!clyde!bonnie!akgua!sdcsvax!sdcrdcf!hplabs!hao!seismo!brl-tgr!tgr!LARRY@JPL-VLSI.ARPA From: Larry CarrollNewsgroups: net.unix Subject: RatFor on the IBM PC/AT Message-ID: <6821@brl-tgr.ARPA> Date: Fri, 28-Dec-84 19:16:30 EST Article-I.D.: brl-tgr.6821 Posted: Fri Dec 28 19:16:30 1984 Date-Received: Sun, 30-Dec-84 01:58:33 EST Sender: news@brl-tgr.ARPA Organization: Ballistic Research Lab Lines: 20 You must have been told wrong. The purpose of RatFor is to allow you to write C-like code which is translated into ForTran source for compilation by a ForTran compiler. It's a public domain program for those with no (or a terrible) C compiler who do have a ForTran compiler. The ForTran source produced by RatFor (and other such interpreters) is pretty ugly and introduces a 5-10% speed penalty. Also, the ForTran compilers occasionally do some erroneous optimization of RatFor'd code. These errors are pretty rare, and the speed penalty is generally made up for by the excellence of ForTran compilers. The ugly ForTran is looked at only when you're looking at a hex/octal dump to find the exact location and nature of an error. Then you change the RatFor code, not the ForTran code. I've a question of my own about Unix/Xenix on the AT. Benchmarks have shown only a 2-3 times speed advantage of the AT over the PC. If you use a multi-user OS, doesn't the overhead pretty much eat up that advantage, giving you in effect only PC capability? Larry @ jpl-vlsi ------