[ExI] petaflop computers

Damien Broderick thespike at satx.rr.com
Thu Nov 20 04:57:34 UTC 2008


At 07:36 PM 11/19/2008 -0800, Spike Jones wrote:

>They keep calling it a petaflop barrier, but I know of no barrier.  Still,
>it's cool that it was exceeded:
>
>http://blog.wired.com/wiredscience/2008/11/supercomputers.html

Here's what I wrote some 8 years ago in the US edition of THE SPIKE, 
when the first machine had just managed teraflop computing:

<That is a scary jump in power. We are talking about an array of 
machines doing a million million calculations each second. When it 
was finished in 1997, it sustained 1.4 teraflops and peaked at about 
two teraflops. Not to be outdone, late in 1999, IBM signed a $100 
million contract with the American government to build an 
ultrasupercomputer within five years. Named Blue Gene, a descendent 
of the Deep Blue machine that unnerved chess grandmaster Gary 
Kasparov in 1997, it would run a *thousand* trillion operations a 
second (a petaflop), two million times faster than today's best 
desktop computers.
           It is easy to see that Moore's Law is right on target for 
the Spike [i.e. the Vingean technological singularity, for any newbies]. >

Well, it seems to have taken a little longer than 5 years, but hey--a 
thousand times faster is 10 years of annual doubling. About on the money.

Damien Broderick 




More information about the extropy-chat mailing list