BENCHMARKS (Byte's) 368 vs 486 comp.

Bob Palowoda palowoda at fiver.UUCP
Wed Mar 14 18:47:22 AEST 1990


>From article <36 at cimcor.MN.ORG>, by mike at cimcor.MN.ORG (Michael Grenier):
> From article <1025 at fiver.UUCP>, by palowoda at fiver.UUCP (Bob Palowoda):
>>   You are right. Here are the revised numbers.
>>   Once again 25Mhz 386, pcc.
>> --------------------------------------------------------------------------
>> Without reg
>> Dhrystones/sec:               5400             5400            141.14
> 
> 
> The dhrystones numbers here don't seem right either. There must be
> a lot of waits states on the memory or something. Even my lowly 20 Mhz
> 386 without memory interleaving hits 6200 dhrystones (without 
> registers) and my brother's 25Mhz 386 with 8 Meg of ram allowing for
> interleaving regularly hits 10,000 dhrystones.

 I actually tried this on another 25Mhz system with 10megs. It should
have been doing 4 way interleave. Guess what the Dhr/secs where down
to 2500. I have to verify that the system didn't have any 16bit memory
up in the last two meg. I notice if you put any 16bit memory in even
the 32bit memory becomes slower. 

 As for hitting the 10000 mark, I don't think I can do it even if I did
go to interleave. My guess is I would only get a 10% performance increase
with interleave and that would leave me with just about the result you 
have on your 20Mhz system. So something else has to be changeing the
numbers. I'm curious do you have any networking software install? Such
as network daemons.  

 The numbers on 486 are nice, you can actually feel the difference when
useing the system. The only thing that's got me baffled is why the 
variance is so high? Maybe it was in cache all the time.

---Bob


> 
> Still the numbers that were posted for the 486 are impressive.
> 
>      -Mike Grenier
>      mike at cimcor.mn.org
>      uunet!rosevax!cimcor!mike
-- 
Bob Palowoda  indetech!fiver!palowoda      *Home of Fiver BBS*  login: bbs
Home {sun|daisy}!ys2!fiver!palowoda            (415)-623-8809 1200/2400
Work {sun|pyramid|decwrl}!megatest!palowoda (415)-623-8806 2400/9600/19200 TB
Voice: (415)-623-7495  palowoda at fiver           Public access UNIX XBBS   



More information about the Comp.unix.i386 mailing list