[RC5] Did the calculations again...

Elektron elektron_rc5 at yahoo.ca
Wed Jan 7 01:25:51 EST 2004


On 7 Jan, 2004, at 05:41, Daniel Quintiliani wrote:

> 163536894738 Keys/s produces
> k = 5.160831909 * 10^18

http://www1.distributed.net/~pstadt/rc5-72/

First, it shows that the average keyrate now is closer to 130 G (it may 
be more like 120, because I tink the recent peaks are due to the 
downtime in stats or something).

It also shows that keyrate doesn't increase exponentially with time 
(not yet anyway).

> Our formula is:
> t = ln(2^72/((5.160831909 * 10^18) ln(2^(1/1.5)))) / ln(2^(1/1.5))
>
> That comes to 16.42712929 years (16 years, beginning of June) for all 
> of the keys to be finished (from 399 days ago). That's a 9 month 
> reduction from a calculation three months ago! We may finish much 
> sooner than we all think, especially since we usually find the key 
> before we reach 100%.
>
> Now I see what you mean about new participants. We should add that to 
> the stats - when 100% will be reached if Moore's Law holds true during 
> that time if no new participants join. (We currently only have those 
> stats *without* calculating Moore's Law.) Check my math once more and 
> I'll submit an enhancement request.

Apparently the old calculation gave 17.20178906 years

With 160 Gkeys/s, I get 16.92378607786446985959.

With 2^72*(1-0.00068) total keys (because .068% is done), I get 
16.92231402836919245458.

That's a difference of 0.00147204949527740501 years, or 
12.90398587560173231766 hours.

So all the work on RC5-72 so far amounts to half a day (of course, this 
is because the extra keys, in 16 years' time, would take half a day to 
do).

- Purr


END



More information about the rc5 mailing list