Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!uwvax!oddjob!gargoyle!ihnp4!chinet!nucsrl!ram
From: ram@nucsrl.UUCP (Renu Raman)
Newsgroups: comp.arch
Subject: Re: Question:  on-chip or off-chip MMU?
Message-ID: <3810037@nucsrl.UUCP>
Date: Tue, 2-Jun-87 12:33:34 EDT
Article-I.D.: nucsrl.3810037
Posted: Tue Jun  2 12:33:34 1987
Date-Received: Sat, 6-Jun-87 20:41:38 EDT
References: <579@gec-mi-at.co.uk>
Organization: Northwestern U, Evanston IL, USA
Lines: 34

Spencer wrote:

>>      Another nice thing about optical computing is the ability to represent
>>      and compute in multiple levels of logic.  Imagine ternary or other
>>      higher radices used for computation...

>I am thinking about it.  I'm wondering "why bother?".  Binary works really
>well, and puts the least demands on linearity and reproducibility of device
>properties -- *not* a strong point of optical digital devices to date.

	Info of interest:  The iAPX 432 and I believe the 8087 (source:
	A friend over here who works on MVL - Multiple Valued Logic)
	had their microcode ROM based on MVL.  That enabled a reduction
	in chip area by 30-40% for the ROM part of the chip although
	imposing a penalty on time due to additional levels of interpretation.
	[Also, I believe there are a number of PLAs and/or memory cells
	based on MVL being used commercially]

>>    If optical computing becomes commonplace, what would be the equivalent of
>>    MIPS and FLOPS? MQPS (Million Quanta per second :-))

>MIPS and FLOPS, basically.  I don't *care* how quickly you can do a Fourier
>transform, that won't compile my program.

	OK MIPS will compile. But MFLOPS?
>-- 
>"The average nutritional value    Henry Spencer @ U of Toronto Zoology
>of promises is roughly zero."     {allegra,ihnp4,decvax,pyramid}!utzoo!henry
>----------

-------------------
Renu Raman				UUCP:...ihnp4!nucsrl!ram
1410 Chicago Ave., #505			ARPA:ram@eecs.nwu.edu
Evanston  IL  60201			AT&T:(312)-869-4276