GPU72 plans post-announcement
 

After the new Mersenne prime is announced there is likely to be a surge of new users. This has had the effect of seeding up the LL testing wave in the past. What are GPU72's plans to deal with this surge? Are we going to release exponents at a lower bit level? It makes no sense to hold on to 61Ms to go to 73 when 62Ms are likely to be handed out done only to 70. Should we try a BFS approach for a short period bringing in say everything to 63M?

What does George think? Keep in mind that many of those reservations will end up being abandoned.

Whether or not there is a surge in new users is totally dependent on the amount of press coverage which varies widely with each new prime. I'd be surprised if any surge would significantly eat into the lead GPU72 has on the LL wavefront. If it does, then we can react accordingly.

[QUOTE=garo;327510]What are GPU72's plans to deal with this surge?[/QUOTE]

We have our day clear to deal with, in real-time, the situation.

Never send a human to do a machine's job. A corollary: never send a machine to do a human's job.

I am not talking about a surge on the day. It is the increase in the rate of reservations that lasts for about a month after an announcement. But you are right, let the machine do its job.

PS: What is the lead of GPU72 over the LL wave front?
PPS: Scratch that. I see about 33k exponents at 73 bits in the 60-64M range.

[QUOTE=chalsall;327539]We have our day clear to deal with, in real-time, the situation.[/QUOTE]

Unless /. posts a blurb, you might not even see a blip.

[QUOTE=Prime95;327547]Unless /. posts a blurb, you might not even see a blip.[/QUOTE]

Yes, I'm paranoid. The question is am I paranoid enough.

[QUOTE=chalsall;327548]Yes, I'm paranoid. The question is am I paranoid enough.[/QUOTE]


GPU72 does have a check to prevent someone from overdrawing on their work requests, and with its 30 day expiration on active requests the number of abandons may not cause significant disruption of the actual work metrics.

[QUOTE=swl551;327551]GPU72 does have a check to prevent someone from overdrawing on their work requests, and with its 30 day expiration on active requests the number of abandons may not cause significant disruption of the actual work metrics.[/QUOTE]

But the masses are not going to be using GPU72, they will get work directly from GIMPS.

[QUOTE=Chuck;327598]But the masses are not going to be using GPU72, they will get work directly from GIMPS.[/QUOTE]

No, but I am hopeful that we'll get some more GPU participants as a result of the announcement. And I also have to monitor [url]www.mersenne.info[/url].

It should all be fine, but having been slashdot'ed myself more than once, I know the traffic load can be extreme for a while....

[QUOTE=chalsall;327713]
It should all be fine, but having been slashdot'ed myself more than once, I know the traffic load can be extreme for a while....[/QUOTE]

I don't know what that means.

[QUOTE=Chuck;327732]I don't know what that means.[/QUOTE]

[URL="http://en.wikipedia.org/wiki/Slashdot_effect"]The Slashdot effect[/URL]....

Naturally, George and Scott have a lot more to worry about. But Mike, James and I might have to deal with some unusual load ourselves.