GPU to 72 status...

[petrw1]

Originally Posted by chalsall

Oh, and just for the record, Jon Pace is continuing to contribute ~3.5 THzD/D of work.

My bad

[James Heinrich]

For what it's worth, my tf1G project throughput seems to have tanked since Dec 20 (down by ~90%), so perhaps those who were applying resources there have redirected them to more (immediately) useful work such as is being discussed here. And I'm perfectly fine with that.

[chalsall]

Originally Posted by chalsall

One thing which might be helpful is if Primenet would assign work sorted by TF level desc (as with P-1, grouped by 1M range). That way GPU72 workers could focus on the high end of each range, as the wavefronts race towards us.

George: your thoughts on specifically this suggestion?

In my mind, there is little sense in GPU72 issuing work in 10xM unless this delta is done on Primenet. Otherwise, there is little hope in having work done even to 74 "bits" and having only that picked up by LL'ers without reserving everything in a 1M range.

Having GPU72 reserve everything would prevent Primenet from issuing TF'ing work itself which would then soon be issued to LL'ers; IMO, a bad idea. There are several people working directly with Primenet -- the two "workforces" should be cooperating here. GPU72 does some work at the high end of each 1M range; Primenet does the same.

In the short form: Primenet should "order by [TF Level] desc, [P1 done] desc group by [1M range]". My understanding is currently only the latter clause is in the query, which no longer has any real impact since almost no P-1'ing has been done at 100M and above.

Thoughts?

[petrw1]

Originally Posted by chalsall

George: your thoughts on specifically this suggestion?

In my mind, there is little sense in GPU72 issuing work in 10xM unless this delta is done on Primenet. Otherwise, there is little hope in having work done even to 74 "bits" and having only that picked up by LL'ers without reserving everything in a 1M range.

Having GPU72 reserve everything would prevent Primenet from issuing TF'ing work itself which would then soon be issued to LL'ers; IMO, a bad idea. There are several people working directly with Primenet -- the two "workforces" should be cooperating here. GPU72 does some work at the high end of each 1M range; Primenet does the same.

In the short form: Primenet should "order by [TF Level] desc, [P1 done] desc group by [1M range]". My understanding is currently only the latter clause is in the query, which no longer has any real impact since almost no P-1'ing has been done at 100M and above.

Thoughts?

As I understand The Math: https://www.mersenne.org/various/math.php
specifically on the topic of TF, PrimeNet has already taken all exponents up to 190M to the prescribed limits;
that is, 1 bit below "The Math" page which then needs a P-1 before the last bit.

That said, I agree that a coordinated effort between GPU72 and PrimeNet is the best offense;
but I suspect for that to happen George would need to change the PrimeNet TF limit rules.

[chalsall]

Originally Posted by petrw1

As I understand The Math: https://www.mersenne.org/various/math.php specifically on the topic of TF, PrimeNet has already taken all exponents up to 190M to the prescribed limits;

Those maths are from before GPUs entered the equation.

Originally Posted by petrw1

...but I suspect for that to happen George would need to change the PrimeNet TF limit rules.

Nope. It would only be one (or only a few) changes to SQL statement(s). Include the "TF Depth" clause into the SQL query / sort statement(s). No other changes are needed.

[James Heinrich]

Originally Posted by petrw1

As I understand The Math...
PrimeNet has already taken all exponents up to 190M to the prescribed limits

Remember those limits are at least 10+ years old and derived from the before-time, before GPU-TF existed. Those numbers are no longer accurate (by a large margin).
Note on my graphs, the red curved line is the old PrimeNet TF limits as per the Math page, the purple line is PrimeNet+3 bits (old GPU72 target), the greenish line is PrimeNet+5 bits (new GPU72 target).

[Prime95]

Originally Posted by chalsall

George: your thoughts on specifically this suggestion?

I think I can do that -- if I could log on to the server.

[chalsall]

Originally Posted by Prime95

I think I can do that -- if I could log on to the server.

A fairly trivial delta. Do you not have support staff?

P.S. I find I have to be *very* careful in what I say in some spaces. The immediate above was meant to be both funny, and serious, at the same time.

[Prime95]

Originally Posted by chalsall

Do you not have support staff?

More unpaid volunteers

[Prime95]

Originally Posted by chalsall

One thing which might be helpful is if Primenet would assign work sorted by TF level desc (as with P-1, grouped by 1M range). That way GPU72 workers could focus on the high end of each range, as the wavefronts race towards us

I think this is done.

It may operate more slowly as I used to have an index that matched the sort order, now I think SQLServer will need to scan the available exponents in a 1M range.

Let me know if I screwed up.

[storm5510]

I periodically check my assignments on the GPU72 web site. Today, I found five which were close to expiration. When I click on each number, another page appears from mersenne.org which shows I have completed the assignment. See the attached image below.

The remainder, which are not highlighted, I have queued in mfaktc to get them done. This includes one Colaboratory assignment. One of my instances completed a single exponent this past Sunday, then stopped part way through a second. I need to look at my assignments page much more often.