Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!ames!apple!usc!ucla-cs!marc
From: marc@oahu.cs.ucla.edu (Marc Tremblay)
Newsgroups: comp.arch
Subject: Re: HOT CHIPS conference
Message-ID: <25359@shemp.CS.UCLA.EDU>
Date: 30 Jun 89 16:26:15 GMT
References: <596@megatek.UUCP> <112807@sun.Eng.Sun.COM> <142@ssp1.idca.tds.philips.nl> <113277@sun.Eng.Sun.COM> <145@ssp1.idca.tds.philips.nl>
Sender: news@CS.UCLA.EDU
Reply-To: marc@cs.ucla.edu (Marc Tremblay)
Organization: UCLA Computer Science Department
Lines: 29

>>	NexGen (really,really fast 386 multichip solution ... not 
>>	        fully silicon yet)
>>
>Will the i486 kill the NexGen?
>
>And these:
>
>Whats the estimated performance differential between the two?

David Stiles, who gave an excellent presentation, 
estimated the performance of the NexGen chip set to be almost
twice as fast as the 486. If I remember correctly he had an
extra slide showing that: 

		NexGen = 1.9 X i486

There are lots of nice features in their architecture, such
as out-of-order completion, virtual register assignment,
write reservation tables, etc. All those features are
implemented using about 2.2 million transistors spreaded on 8 chips.
One of the busses runs at 66 MHz making the synchronization between 
the chips quite tricky (they use phase-locked-loop).

The chip set represents a lot of hard work and several good ideas but 
I wonder if they had anticipated that the i486 would be as fast as 
it turned out to be.

					Marc Tremblay
					marc@CS.UCLA.EDU