Path: utzoo!attcan!utgpu!cunews!bnrgate!bigsur!bnr-rsc!bcarh185!schow
From: schow@bcarh185.bnr.ca (Stanley T.H. Chow)
Newsgroups: comp.misc
Subject: Re: Accessory to double speed of 486 PCs
Message-ID: <3843@bnr-rsc.UUCP>
Date: 18 Dec 90 18:45:24 GMT
References: <2370004@hpwrce.HP.COM> <2370005@hpwrce.HP.COM>
Sender: news@bnr-rsc.UUCP
Reply-To: bcarh185!schow@bnr-rsc.UUCP (Stanley T.H. Chow)
Organization: BNR Ottawa, Canada
Lines: 29
Summary:
Followup-To:
Keywords:

In article <2370005@hpwrce.HP.COM> kingsley@hpwrce.HP.COM (Kingsley Morse) writes:
>schow@bcarh185.bnr.ca (Stanley T.H. Chow) writes:
>
>>Getting a motherboard running at 50 MHz is a lot more difficult
>
>Would it be possible to keep the motherboard at 25 MHz and rely on speeding
>up the cache on the cpu chip to get some speed increase for small (or local) 
>programs?

Yes, it is certainly possible. That is how the Everex machine was set up.
We also do it in some of our processors in telephone switches. The problem,
is of course, Cache Hit-rate. 

The actual speed up will vary depending on the cache size, cache fill
algorithm, cache replacement algorithm, delays caused by interfacing at two
speeds, etc. For example, in our product, going from a 68030 at 20 MHz to a
68030 at 40 MHz (with 20 MHz memory, but 512K of 40MHz SRAM) got a factor
of 1.5 speed up instead of a factor of two.

For most PC compute-intensive applications, the bare i486 runing at double
speed is likely to get only a factor of 1.2 speed up. With more cache at
50 MHz, the speed up may get as high as 1.5. On the otherhand, going from
25MHz to 33MHz will get you a factor of 1.3.


Stanley Chow        BitNet:  schow@BNR.CA
BNR		    UUCP:    ..!uunet!bnrgate!bcarh185!schow
(613) 763-2831               ..!psuvax1!BNR.CA.bitnet!schow
Me? Represent other people? Don't make them laugh so hard.