Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!rphroy!caen!uwm.edu!psuvax1!rutgers!rochester!pt.cs.cmu.edu!gandalf.cs.cmu.edu!lindsay
From: lindsay@gandalf.cs.cmu.edu (Donald Lindsay)
Newsgroups: comp.arch
Subject: Re: iWarp Architecture
Message-ID: <13338@pt.cs.cmu.edu>
Date: 9 Jun 91 01:59:17 GMT
References: <1991Jun3.172230.6901@iWarp.intel.com> <2622@m1.cs.man.ac.uk> <4077@ssc-bee.ssc-vax.UUCP>
Organization: Carnegie Mellon
Lines: 24

In article <4077@ssc-bee.ssc-vax.UUCP> 
	carroll@ssc-vax.UUCP (Jeff Carroll) writes:
>Professor Kung has been pushing the Warp project at CMU for
>a decade or so now, and Intel has been at work on iWarp for several years
>funded in part by the Department of Defense.

About right. The first Warp was wire wrapped: the second was PCB: and
now, by collaboration with Intel, Kung has got a node down to one
chip + RAM chips. The design has evolved somewhat in that time, with
the compiler people having a large say. The communications issues
have received a lot of study, and there is (gasp!) an actual user
community for the pre-intel Warp, albeit a very small one. 

[Just last week, my smiling face was image-processed through a Warp:
this in aid of a local enhancement to the idea of "finger"ing people.
Kung has in fact taken the I/O issue quite seriously, and is a major
force behind our efforts in high speed networking.]

This isn't Intel's only collaboration. They are building chips for
Dally's J-machine, for example. This strikes me as quite intelligent:
costly, but it brings new experiences into the company, and any one
project just might wind up paying for all of them.
-- 
Don		D.C.Lindsay 	Carnegie Mellon Robotics Institute