Re: Notes on packing serial cores into Xilinx devices

jbass-/[email protected] Sun, 15 Aug 2004 12:46:14 -0600
Newsgroups gmane.comp.distributed-net.hardware
Message-ID <[email protected]>
Elektron <elektron_rc5-FFYn/[email protected]> writes:
> Now, you win $10000. Assume electricity costs an insanely low 1 cent 
> per kWh. Assume equipment is free. So you have 1 million kWh (== 3.6e12 
> J) to spend cracking 2^71 keys (since, on average, you search half the 
> keyspace to find the result). That's 1.5e-9 Joules per key.

This argument only makes sense if some central admin is making a reasoned
choice to participate in the project. From an individul perspective, at
$4/mo to leave a PC running with lottery like odds to potentially win
$10,000 ... the breakeven point for a winner is 10000/4= 2,500 months.

Lotteries don't make sense from a global systems perspective, as the
operator will always win. Similar game theory results on this problem.

> Let's estimate the minimum number of transistor*clocks you need to do 
> RC5. Assume you need no bits of storage (which you don't if you're not 
> pipelining). Skip this if you don't feel like counting transistors.

Sorry, but 26 32bit words of S storage are ALWAYS required, as are 3
32 bit words of L.

John
_______________________________________________
Hardware mailing list
Hardware-Ra3b/QYEcJ3d140v2zMXi0fjHoOT/h/[email protected]
http://lists.distributed.net/mailman/listinfo/hardware