[RC5] power consumption

Thorsten Wolf t_wolf at gmx.de
Tue Jul 3 01:48:19 EDT 2012


Dear Paul.

You've given your message some thought, so I will give my reply some as well.

First, I think it is right that this kind of computing should be done
primarily on "up-to-date" hardware, which by tech advantages will be
more efficient in terms of calculations done per consumed Watts....

But:

People choose to participate in distributed computing. Distributed.net
is one of the first distributed computing projects out there, but if
you check BOINC, you'll see it's not the last.
People choose to participate because they believe in the goal of the
project they participate in or they're just hungry for the credits.
If Distributed.net closes for let's say 36 months, people will still
have GPUs and CPUs running at load. It'll just be different projects.

Also, I don't believe that "Hey, let's wait for something better to
come up and do our work in a few years..." ever got anyone any further
to their goal.

I've been with d.net since many years and though I don't crunch
recently, I still believe that halting the project would make any
change. And I'm surely not going to start a flame war or global
environment discussion.

Thanks for your time and effort you put into your message. If anything
should be discussed about, it could be stopping CPUs and having only
GPUs compute (since they are way more efficient)... But that's not
mine to decide anyways.

Cheers,

Thorsten

On Mon, Jul 2, 2012 at 11:37 PM, Relkin, Paul <pwrelkin at mail.roanoke.edu> wrote:
> A 2012 Radeon HD 7970 GPU uses 215 additional watts when crunching RC5-72.
> Completing the rest of RC5-72 using only Radeon HD 7970s would consume 77
> million kilowatt hours of electricity. If participants pay 10 cents per
> kilowatt hour, they would be donating $7.7 million of electricity. This
> would emit about 100 million pounds of carbon dioxide, as much as 5 million
> gallons of gas. Currently, most distributed.net clients are running on
> hardware that is less efficient than that.
>
> Until 2009, RC5-72 ran entirely on CPUs rather than GPUs. In the first 6
> years of RC5-72 (2002 to 2008), distributed.net caused thousands of CPUs to
> run at maximum power. The progress made by these CPUs is barely significant.
> The total keyspace completed before 2009 could now be processed in the same
> amount of time by only 38 Radeon HD 7970 GPUs, saving a lot of electricity.
>
> I have participated in distributed.net since 2006 but I don't think it is
> ethical to continue crunching without evaluating the environmental impact.
> Processing speed and energy efficiency are improving exponentially. If
> distributed.net was paused for a few years, the cost to participants and the
> environment could be cut exponentially.
>
> It's exciting that it is becoming increasingly feasible to crack RC5-72. But
> for now, I'm not sure that cracking the secret message a few years sooner is
> worth the environmental and economic consequences of crunching on the less
> efficient hardware available today.
>
> Paul Relkin
>
> _______________________________________________
> rc5 mailing list
> rc5 at lists.distributed.net
> http://lists.distributed.net/mailman/listinfo/rc5
>


More information about the rc5 mailing list