GPU to 72 status...

[chalsall]

Originally Posted by ixfd64

I noticed GPU to 72 is no longer giving assignments to 73 bits. Any idea when 73-bit assignments will be available again?

Never. Please see this post and onwards.

[ixfd64]

Originally Posted by chalsall

Never. Please see this post and onwards.

I was under the impression that this only applied to exponents in the 94m range.

There are actually still many exponents that have only been factored to 72 bits. As for the person working "off the books," I don't see anyone submitting large amounts of TF results aside from the usual names. Is there something I'm missing?

[chalsall]

Originally Posted by ixfd64

There are actually still many exponents that have only been factored to 72 bits.

Yes, but these are in 98M and above. Not going to be needed for many years. And, again, some have been working these "off-the-books", stepping on toes.

Originally Posted by ixfd64

Is there something I'm missing?

I don't think so. But, again, GPU72 was created to help the GIMPS project.

If you want to do TF'ing to 73 bits up there, you're free to get them directly from Primenet.

[James Heinrich]

Originally Posted by chalsall

Yes, but these are in 98M and above. Not going to be needed for many years

As a pointless point of comparison, in the 1000M-4294M range, there are 72,698,150 candidates still needing TF to 2⁷³, and at the current workrate that's estimated to take 46.4 years to complete. And by the time the end of that range is actually useful for GIMPs, Andromeda will be set in the past.
More likely some grand computing breakthrough will happen long before then and that kind of processing will become trivial.
I have no idea who would work in such a range...

[ixfd64]

Originally Posted by chalsall

And, again, some have been working these "off-the-books", stepping on toes.

On second thought, is Niels_Mache_Nextcloud the person you're talking about? He's been working on many exponents in the 98m and 99m ranges, neither of which are currently assigned by GPU to 72 or PrimeNet.

[chalsall]

Originally Posted by ixfd64

On second thought, is Niels_Mache_Nextcloud the person you're talking about? He's been working on many exponents in the 98m and 99m ranges, neither of which are currently assigned by GPU to 72 or PrimeNet.

I didn't want to "name and shame", but yes.

Even though GPU72 had reserved the work in 94M from Primenet and "lent" it out to participants, he still submitted his results to Primenet.

[ixfd64]

I see. Someone might want to gently inform him that GPU to 72 and the manual GPU assignments page are a thing. He might not know that they exist as neither are linked anywhere from the official GIMPS website.

[James Heinrich]

Originally Posted by ixfd64

He might not know that they exist as neither are linked anywhere from the official GIMPS website.

I have added a small note of the top of each page and cross-linked them.

[ixfd64]

I contacted Niels on Facebook several days ago but haven't gotten a response. It's possible he hasn't even seen my message due to Facebook's overzealous spam filters.

[ixfd64]

I just heard back from Niels. He told me the reason for working "off the books" is that he was running into restrictions when reserving exponents. However, he has agreed to reserve the ranges beforehand now that he's aware of GPU to 72 and the manual GPU assignments page. Hopefully we can avoid stepping on toes from now on.

[Chuck]

I see that our anonymous factorer is back at work — 32,300+ current assignments. S/he likes round numbers in the "Workers' Overall Progress" table and I see that 73 bits for this user has a non-round value.

Looking at the available TF work, it looks like this person is taking all available candidates to 73 bits which have already been double-checked (clearing out the 38M to 43M ranges).