Path: utzoo!attcan!uunet!lll-winken!lll-lcc!ames!ncar!tank!uxc!uxc.cso.uiuc.edu!mcdurb!aglew
From: aglew@mcdurb.Urbana.Gould.COM
Newsgroups: comp.arch
Subject: Re: Content Addressible Memories
Message-ID: <28200247@mcdurb>
Date: 13 Dec 88 17:48:00 GMT
References: <367@enint.Wichita.NCR.COM>
Lines: 19
Nf-ID: #R:enint.Wichita.NCR.COM:367:mcdurb:28200247:000:926
Nf-From: mcdurb.Urbana.Gould.COM!aglew    Dec 13 11:48:00 1988


>"The industry's first VLSI CAM is now available from AMD. The
>high-performance Am99C10, with a capacity of 256 words and a
>user-programmable word width of 16 bits or 48 bits, features a totally
>new concept in memory devices -- memory with built-in logic on every
>cell. 

I think this press release expresses the basic problem with
CAMs - they are too small. For the tasks that really need CAM
type parallelism, 256 words is far too few. To process really
large data sets this way you will have to load the CAM, match,
load the CAM again, match, and so on...

However, it's a start. AMD's CAM will be useful for applications 
that can afford to put enough of them together to make a real system,
as well as for special purpose applications, like searching an
entire track of a disk for a given pattern. 

Out of curiousity, what other applications are people buying the
AMD CAM for (as opposed to wanting to apply it to)?