Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!utgpu!water!watmath!clyde!cbatt!ucbvax!jade!lapis.berkeley.edu!mic From: mic@lapis.berkeley.edu.UUCP Newsgroups: comp.sys.ibm.pc Subject: NUMBERCRUNCHING on the PC: Should C be used at all? Message-ID: <3254@jade.BERKELEY.EDU> Date: Tue, 21-Apr-87 18:17:20 EST Article-I.D.: jade.3254 Posted: Tue Apr 21 18:17:20 1987 Date-Received: Thu, 23-Apr-87 23:36:24 EST Sender: usenet@jade.BERKELEY.EDU Reply-To: mic@lapis.berkeley.edu(Michel Bruneau) Organization: University of California, Berkeley Lines: 26 In reply to my previous question, I am convinced of the vertues of C for getting a closeness to the system without having to mess-up with the assembler. But my original question was not well formulated (I forgot I was talking to system programmers here). NOW: In an environment where 95% of the work to be done by a program is number-crunching, (talking about structural analysis for civil engineerings), most people as using.. well..FORTRAN (That's it I said the dreaded word, FORTRAN the archaic!). Anyhow, as of today, I am not convinced that something like C will help me do the job faster (key word), although I realise all af of the nice tricks C will allow me to do. C would maybe help me to get some nice friendly features implemented, but not all: For example, engineers like standards (why would you think they stuck to Fortran for all those years). So when come time to add graphics to a structural analysis program, it would appear logical to use G.K.S. (another clanky standard). Unfortunately, I have not heard of a single C compiler that will link with the GKS librairy (but 3 fortrans do). Further, graphics are frill, speed is the essence of those application (i.e. number-crunching speed). Not everybody wants to buy a CRAY, but one is entitled to the best performance on a given machine. THEREFORE, as anybody seen number crunching benchmarks for pure calculation (no I/O) comparing C and FORTRAN???? I am still ignorant, so please let me know....