Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!ames!lll-winken!uunet!cbmvax!jesup
From: jesup@cbmvax.UUCP (Randell Jesup)
Newsgroups: comp.arch
Subject: Re: 80486 vs. 68040 code size
Message-ID: <6805@cbmvax.UUCP>
Date: 10 May 89 04:31:39 GMT
References: <907@aber-cs.UUCP> <4103@ficc.uu.net> <9329@b-tech.ann-arbor.mi.us>
Reply-To: jesup@cbmvax.UUCP (Randell Jesup)
Distribution: eunet,world
Organization: Commodore Technology, West Chester, PA
Lines: 21

In article <9329@b-tech.ann-arbor.mi.us> zeeff@b-tech.ann-arbor.mi.us (Jon Zeeff) writes:
>If you take a look at the latest dhrystone (for whatever it's worth) 
>results, you'll see that the 80386 does it faster than the 68030.  

	Dhrystone falls down a bit here, since it over-relies on an unusually
large number of string operations, and the string compares are unusually
long (before a difference).  I believe the x86 family has string instructions
that help it with these operations.

>There seems to be alot of variability though - like a 16Mhz '386 is 
>faster than a NeXt 33Mhz 68030 (both using GCC)?  

	Interesting.  How about this: ~8500 dhrystones (register version)
for a 25Mhz '030 Amiga (gets a bit more if you disable task-switching - it
keeps polling the disk drives for insertion/removal, executing VBlank
interrupts, etc).  One of the many reasons why dhrystones should be taken
with a grain of salt, especially for comparing processors (Compiler tech 
gets in here too - the 8500 was with a microcomputer C compiler).

-- 
Randell Jesup, Commodore Engineering {uunet|rutgers|allegra}!cbmvax!jesup