Return to BSD News archive
Xref: sserve comp.os.linux.misc:464 comp.os.386bsd.misc:728 Newsgroups: comp.os.linux.misc,comp.os.386bsd.misc Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!agate!spool.mu.edu!torn!nott!bnrgate!bmerh85!mlord From: mlord@bnr.ca (Mark Lord) Subject: Re: Why would I want LINUX? Message-ID: <1993Aug21.203251.26369@bmerh85.bnr.ca> Sender: news@bmerh85.bnr.ca (Usenet News) Organization: BNR Ottawa, Canada References: <MIKE.93Aug19115915@pdx800.jf.intel.com> <250m5t$dmk@europa.eng.gtefsd.com> <252n71$2d4@fnnews.fnal.gov> Date: Sat, 21 Aug 93 20:32:51 GMT Lines: 18 In article <252n71$2d4@fnnews.fnal.gov> dejan@cdfsga.fnal.gov writes: ... > Now, the copmarison. Those were EXACTLY THE SAME MACHINES. Bought from >a same vendor, exactly the same equipment inside, 387 FPU in both of them. >Fortran on DOS was an expensive commercial product, it was dos 5.0 if I >remember well, and under DOS the program ran about a minute and five seconds >on both of them. We ran the program on BSD, fifteen seconds. Well, I know >that in real mode 386 emulates 32bit integer operations, but FOUR TIMES >FASTER!? Get real! Most of this speedup is very likely due to the quality of the 386-specific C compiler used on BSD. But then, that *is* a valid thing to include in many such comparisms. However, were the program built for MSDOS with a compiler that generated code of similar quality, the two run-times (excluding any IO) would then compare much more closely. With a decent MSDOS cache (HYPERDISK) even the I/O would be as good for DOS as for BSD (if not better). -- mlord@bnr.ca Mark Lord BNR Ottawa,Canada 613-763-7482