Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!henry
From: henry@utzoo.UUCP (Henry Spencer)
Newsgroups: net.arch
Subject: Re: "The Shared Memory Hypercube"  Do you smell any smoke?
Message-ID: <5578@utzoo.UUCP>
Date: Thu, 9-May-85 13:59:39 EDT
Article-I.D.: utzoo.5578
Posted: Thu May  9 13:59:39 1985
Date-Received: Thu, 9-May-85 13:59:39 EDT
References: <2132@sun.uucp> <1447@think.ARPA> <551@lll-crg.ARPA>, <973@ames.UUCP>
Organization: U of Toronto Zoology
Lines: 18

> Short vector startup time is very good.  The vector registers are faster
> than the scalar registers after you pass 3 elements.  ...
> ...  DON'T WORRY about
> vector STARTUP on a CRAY, the difference is insignificant, you are
> wasting your time optimizing this arena.  ...

It should be noted that this is probably the single biggest reason why
the Cray machines have been so successful, compared to most previous
supercomputer efforts:  they're scorching fast even on scalars, and get
full vector performance on even short vectors.  This is why sales of
Cray machines are already an order of magnitude larger than the combined
sales of the Star-100 and the ASC, both of which were Real Fast if they
could ever get their pipelines full...  Life is much easier for the poor
users when they don't need to sweat blood to make their vectors long
enough for full-speed operation.
-- 
				Henry Spencer @ U of Toronto Zoology
				{allegra,ihnp4,linus,decvax}!utzoo!henry