Require P-1 and other factoring work to be done on reliable machines

[Madpoo]

Originally Posted by tha

I do care about Mersenne primes. Recently, I even switched one machine to doing three first time LL tests. But I care about factors too. Even if a candidate exponent is proven composite, I like it if one or more factors are known. Just for the sake of it. And I dislike it if the database contains a wrong entry 'no factors below' or 'none within boundaries B1/B2' when that does not happen to be the case.

Yeah, I think there are folks who like things nice and tidy. I don't like gaps in the data either, or the unscientific principle of people checking their own work, etc.

But then there are people who don't seem to mind a little sloppiness around the edges, but I don't have a problem with them. In a group effort there are going to be people who care more about different things and I like to think it balances out. Gleaners following the harvesters.

Although it does seem a little rude to criticize someone for doing something they enjoy just because they don't enjoy the same thing. I'll leave it at that.

[chalsall]

Originally Posted by Madpoo

Although it does seem a little rude to criticize someone for doing something they enjoy just because they don't enjoy the same thing. I'll leave it at that.

+1.

[R.D. Silverman]

Originally Posted by Madpoo

Although it does seem a little rude to criticize someone for doing something they enjoy just because they don't enjoy the same thing. I'll leave it at that.

It also seems rude to me to complain about missed P-1 factors. So what if some are missed?
Stop obsessing over it.

[chalsall]

Originally Posted by R.D. Silverman

It also seems rude to me to complain about missed P-1 factors. So what if some are missed?

Perhaps those with their own fire-power might enjoy the (possibly very small) chance of optimizing the GIMPS project (very slightly), and possibly, demonstrating empirically that perhaps high error-rate P-1'ing and TF'ing workers might have their previous work re-checked (as is already done for LL'ing)...

We're each here for our own reasons.

Deal with it.

[TObject]

Originally Posted by chalsall

Don't know for sure, but possibly a side-effect of the new assignment rules.

Iff this theory is correct, your machine should start receiving P-1 assignments after two matching DCs are returned.

Well, after two matching double-checks it still grabbed another double-check. Or should it be two matching double-checks per worker?

[TObject]

Scratch that. I just figured it out. It turns out you cannot give 30 GB to Prime95, the maximum you can give it is 29.3 GB. Once I set the memory to 29300 the machine grabbed a P1 assignment.

[Dubslow]

Originally Posted by TObject

Scratch that. I just figured it out. It turns out you cannot give 30 GB to Prime95, the maximum you can give it is 29.3 GB. Once I set the memory to 29300 the machine grabbed a P1 assignment.

[Madpoo]

Originally Posted by TObject

Scratch that. I just figured it out. It turns out you cannot give 30 GB to Prime95, the maximum you can give it is 29.3 GB. Once I set the memory to 29300 the machine grabbed a P1 assignment.

That's slightly puzzling. I don't think there'd be any reason on the client for that... I wonder if the server is making a decision based on what the client says about how much memory it has, and has something going on there.

Good question for George I'm sure. Best I could find is the info on:
http://www.mersenne.org/thresholds/

512MB should be all it needs to get P-1 assignments, but as for why it would stop you if you had more than 30GB setup in either your day or night memory setting, well, that's a mystery. I looked for any clues on that but couldn't see anything at first.

[axn]

IIRC, valid memory setting should be < 90% of available memory. Maybe the client didn't like the "too high" setting and didn't consider it as valid, so server didn't give any P-1 because "invalid memory!"?

[TObject]

It may go of a certain percentage of memory, rather than a fixed amount. But for this particular machine it would not accept more than 29.3 GB. Since I originally tried to give it a little more than that, Prime95 would nit take it.

This also explains why the machhine completed only stage one, when I tried to give it a P1 job manually.

Everything is working fine now.

[Prime95]

Originally Posted by TObject

It may go of a certain percentage of memory.

Yes. The dialog box limits it to something like 85% or 90%, I'd have to look at the code to know. This prevents uninformed users from allocating all their memory and causing severe thrashing. One can override this limitation by directly editing local.txt.