Path: utzoo!attcan!uunet!lll-winken!lll-tis!ames!mailrus!iuvax!isrnix!frank
From: frank@isrnix.UUCP (Frank Burleigh)
Newsgroups: comp.sys.ibm.pc
Subject: Re: MCA
Message-ID: <903@isrnix.UUCP>
Date: 7 Jul 88 14:51:26 GMT
References: <902@isrnix.UUCP> <12400008@cpe>
Reply-To: frank@isrnix.UUCP (Frank Burleigh)
Organization: Inst. for Social Research (Indiana University, 47405)
Lines: 36

In article <12400008@cpe> neese@cpe.UUCP writes:
>
>>
>>on a model 80.  infoworld compared mca memory expansions several issues
>>ago, including the ast advantage/2-386.  they found that board to be equal
>>to the planer board memory speed.  my inquiries to other board makers abt 
>>this claim turn up only denials of any penalty.
>>
>
>Actually, I was not plugging our product, but simply stating the performance
>hits we have noticed with our Model 80-111 and the AST MC Memory Board running
>SCO Xenix.  Regardless of what infoworld stated, there is a penalty for
>accessing memory that occupies an MC slot.  This is by design.  Planar
>memory will always be faster than MC memory.

so, here is a claim.  the engineers may see the logic immediately; i
don't.  please explain how it is "by design" the mca memory is *slower*
(by noticable degrees) than planer memory; how xenix itself might affect
the benchmarks (infoworld used dos and dos applications for their
tests); why it is reasonable to say "regardless of what infoworld says";
whether infoworld's use of the 16mhz model 80 instead of the 20mhz model
80 conditions their tests of the ast board; why 386^max' memory speed
test doesn't reveal this 400% "performance hit" (or any, qualitas says).

i am not personally interested in whether mca is *faster* for existing
applications.  i am more interested in the claims that it is *slower*.

>
>						Roy Neese
>					UUCP @	ihnp4!sys1!cpe!neese


-- 
 Frank C. Burleigh
 Institute for Social Research - Indiana University - Bloomington, IN 47405
 ihnp4!inuxc!isrnix!frank
 {pur-ee,kangaro,iuvax}!isrnix!frank