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