[RC5] Time to completion

Kevin van Haaren KvanHaaren at HNTB.com
Wed Nov 26 10:16:02 EST 1997


Computer speeds double about every 3 years.  CPU speeds double about
every year (yes the two are different).  This is almost exclusively a
CPU project so we should see an even faster increase.

PowerPC is supposed to hit 1GHz in 98.

Kevin

> -----Original Message-----
> From:	Chris Arguin [SMTP:Chris.Arguin at unh.edu]
> Sent:	Wednesday, November 26, 1997 9:52 AM
> To:	rc5 at llamas.net
> Subject:	Re: [RC5] Time to completion
> 
> On Tue, 25 Nov 1997, Dave Avery wrote:
> 
> > Cedric Tefft wrote:
> > > 
> > > If I've done my math correctly it looks like at the current rate
> of 9000
> > > MKeys/sec it will take over 64 years to exhaust the keyspace.  In
> order
> > > to complete the task within one year we would need to increase our
> rate
> > > to almost 595,000 MKeys/sec (i.e. 64 times our current rate).
> Even a
> > > 3-year completion target would require an average rate of 195,000
> > > MKeys/sec.  Does it seem likely that we will be able to increase
> our
> > > current rate by such large factors?
> > 
> > thats about right , but a week ago it was over 100 years to
> completion.
> 
> Why do I feel the sudden urge to analyze how quickly the time to solve
> the
> problem is decresing... "In the first week it was going to take a
> millennium, then a week later about 100 years, and after that 64
> years...", "Wow! Long weeks!"
> 
> Lets see, computer speeds double every oh, let's be real conservative
> and
> call it 5 years. So in 5 years we will only have 27 years left.. Of
> course,
> the internet is growing, so we will get more participants..., also
> when
> people upgrade their machines, many (greater than average anyway) will
> keep their old machines chugging away. I'd guess that a maximum of 15
> years is fairly reasonable. Undoubtably, within 6 months we will know
> that
> to be an overestimate :)
> 
> On a side note, I've always imagined a class of problems that could be
> described like this: If you started to work on them now, by the time
> they
> were finished another machine would have been invented that could
> solve
> them before the older machine is done (ie, the sooner the better
> doesn't
> always work). I don't know what to call them, but it's nice to see
> that
> things like distributed.net get around the problem.
> 
> --
> Chris Arguin                 | "I realize, of course, that it's no
> shame to 
> Chris.Aruguin at .unh.edu       |  be poor... But it's no great honor
> either!"
>                              +--------------+	- Fiddler on the Roof
> http://leonardo.sr.unh.edu/arguin/home.html |
> 
> 
> --
> To unsubcribe, send 'unsubscribe rc5' to majordomo at llamas.net
> rc5-digest subscribers replace rc5 with rc5-digest
--
To unsubcribe, send 'unsubscribe rc5' to majordomo at llamas.net
rc5-digest subscribers replace rc5 with rc5-digest



More information about the rc5 mailing list