Path: utzoo!attcan!uunet!husc6!bloom-beacon!think!ames!killer!bigtex!james
From: james@bigtex.uucp (James Van Artsdalen)
Newsgroups: comp.unix.xenix
Subject: Re: CSS Lab Motherboard + SCO Xenix 386
Message-ID: <2091@bigtex.uucp>
Date: 21 May 88 06:36:55 GMT
References: <337@conexch.UUCP> <10858@steinmetz.ge.com>
Reply-To: james@bigtex.UUCP (James Van Artsdalen)
Organization: F.B.N. Software, Austin TX
Lines: 30

IN article <10858@steinmetz.ge.com>, davidsen@crdos1.UUCP (bill davidsen) wrote:
> 1) so far as I know there is one clone board with 64k cache (two if
> Compaq is a clone). It's made by AMI, sole as a board by Mylex, and used
> in a lot of clones because it runs well with 16 bit memory.

PC's Ltd 386/20 uses a cache.  In order to win the horse race with Compaq's
386/20, PCs Ltd uses static column RAM for bulk memory (80ns I think).

> 2) seeing the small improvement of 32k over 64k cache, I'm suspicious
> that there is a point of vanishing returns, and it's not very big.

I have been told that 8K would be perfectly adequate to push the hit rate
well over 90%.

> For information: I ran a benchmark (actually several) comparing an AMI
> (16 MHz version), a Compaq 20 MHz, and the new ALR 20386 (20 HMz with
> The Compaq was about 10% faster CPU than the AMI, and the ALR was 20%
> faster than that. I'm going to wait until the 25 MHz machines are out
> before upgrading.

PC's Ltd designed specifically to beat the Compaq, but I don't know if they
did by 20% or not.  What sort of RAM does the ALR use?

As an aside, the PC's Ltd 286/20 does not use a cache.  My guess is that they
didn't want to compete with their 386/20, since for DOS or OS/2, a 286 is as
good as a 386.  My opinion is that their 386 is the better price/performance
deal.
-- 
James R. Van Artsdalen   ...!ut-sally!utastro!bigtex!james   "Live Free or Die"
Home: 512-346-2444 Work: 328-0282; 110 Wild Basin Rd. Ste #230, Austin TX 78746