home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.m68k
- Path: sparky!uunet!gatech!concert!uvaarpa!murdoch!fermi.clas.Virginia.EDU!jvn
- From: jvn@fermi.clas.Virginia.EDU (Julian V. Noble)
- Subject: Re: is there anyone who has compared m68k and ix86 ?
- Message-ID: <1993Jan21.154340.23424@murdoch.acc.Virginia.EDU>
- Sender: usenet@murdoch.acc.Virginia.EDU
- Organization: University of Virginia
- References: <1jj4umINN318@nic.BelWue.DE>
- Date: Thu, 21 Jan 1993 15:43:40 GMT
- Lines: 16
-
- I have a 68020/68881 Definicon board (16 MHz) in my AST premium computer
- (80286@10Mhz, 80287@8Mhz). The Savage benchmark, in optimized Forth, takes
- 2.5 sec on the 80286/80287 pair, about 5x less on the 68020/68881 pair.
- Clearly, a factor 2x from clock speed. Since the numbers are always on the
- fpu there is no effect from 32-bit vs. 16 bit bus. So we are measuring
- the raw speed difference of 68020/68881 pair vs. 80286/80287 to be 2.5 or
- so. Does this help? The Savage program follows:
-
- TASK SAVAGE
- : (SAVAGE) F=1 F=0 0 DO FOVER F+ F**2 FSQRT FLN FEXP
- FATAN FTAN LOOP ;
- : SAV FINIT ( TIMER ) 2500 (SAVAGE) ( ETIME )
- CR F. FDROP ;
-
- Cheers! --J.V. Noble
-
-