Bypassing unneeded TF assignments

[ATH]

Originally Posted by Uncwilly

Are any missing a PRP or LL?

Yes, it is blocking tons of Cat 1 exponents in 108M and maybe Cat 0 in 107M (haven't checked).

Closer count reveals the true number of SWL551 TF assignments, almost 11,000:

100M-108M: 1346
108M-109M: 9474
109M-110M: 10
I have not checked above 110M.

[kriesel]

Some very slight spot checking indicated adequate TF already, but inadequate or absent P-1.

[Madpoo]

Is the suggestion that we should unreserve those exponents?

I can see they have 10,828 TF assignments, all but 3 of which were assigned on the 22nd.

They also have 4,809 ECM assignments, all assigned on that same day, the 22nd (just after the TF assignments were made).

The most recent results received by the user were back in Feb of 2017 (couple of TF "no factor"). Those were done with mfaktc...

I don't know - it could just be the user fired something up again after a long period away and mistakenly grabbed WAAAY too much. It wouldn't be the first time someone did that.

[Uncwilly]

Originally Posted by Madpoo

Is the suggestion that we should unreserve those exponents?

And moderate how many tests someone can get out of the blue.

[Madpoo]

All of those assignments have now been expired, and the exponents made available again.

[ATH]

Originally Posted by Madpoo

All of those assignments have now been expired, and the exponents made available again.

Thanks.

In my opinion it should not be possible to get TF, P-1, P+1, ECM assignments in Cat 0 + Cat 1 range as well as in the DC Cat 0 + DC Cat 1 range.
Or at least the lower part of the Cat 1 range, if the entire range is too restrictive for the TF/P-1 effort.

[Zhangrc]

Originally Posted by ATH

Thanks.

In my opinion it should not be possible to get TF, P-1, P+1, ECM assignments in Cat 0 + Cat 1 range as well as in the DC Cat 0 + DC Cat 1 range.
Or at least the lower part of the Cat 1 range, if the entire range is too restrictive for the TF/P-1 effort.

I believe that it's OK to do some TF and P-1 in Cat 1 range. For example, 108.3M lies in Cat 1, and so far TF have saved about 19 tests.
what we need is stricter time restrictions, for example, TF must be completed in as little as 20 days, and P-1 must be completed in as little as 10 days.

[S485122]

Originally Posted by Zhangrc

I believe that it's OK to do some TF and P-1 in Cat 1 range. For example, 108.3M lies in Cat 1, and so far TF have saved about 19 tests.
what we need is stricter time restrictions, for example, TF must be completed in as little as 20 days, and P-1 must be completed in as little as 10 days.

A simple hack would be to use the same rules for all type of work in the ranges covered by category rules.
At the same time the first assignment rule could be implemented : "Since PrimeNet began, the server has recycled exponents where the client computer is 60 days past due in updating the server. This rule has been fairly effective and will continue."
There are thousands of assignments that are years over their "Next Update Date".
At the moment it seems as some assignments have a 60 YEARS grace period after their next update date :-)

[chalsall]

Originally Posted by Zhangrc

I believe that it's OK to do some TF and P-1 in Cat 1 range. For example, 108.3M lies in Cat 1, and so far TF have saved about 19 tests. what we need is stricter time restrictions, for example, TF must be completed in as little as 20 days, and P-1 must be completed in as little as 10 days.

Seconded on the value of TF'ing "Just In Time" at the wavefront.

Might I suggest that in Cat 1 (and 0, of course) that Manual assignments aren't allowed? Only "real" assignments (with an AID) that are updated regularly. Expire after, let's say, 10 or 15 days of no update?

[ATH]

Originally Posted by Zhangrc

I believe that it's OK to do some TF and P-1 in Cat 1 range. For example, 108.3M lies in Cat 1, and so far TF have saved about 19 tests.
what we need is stricter time restrictions, for example, TF must be completed in as little as 20 days, and P-1 must be completed in as little as 10 days.

Maybe the first x exponents in Cat 1 should be restricted from TF and P-1 for some appropriate x to be decided. I would say at least x=2000 maybe more.
Or maybe Cat 0 should be expanded a bit, and only Cat 0 restricted from TF/P-1.


https://www.mersenne.org/thresholds/

Category 0
First
200
assignments

Category 1
First
38178
assignments

[chalsall]

Originally Posted by ATH

Maybe the first x exponents in Cat 1 should be restricted from TF and P-1 for some appropriate x to be decided. I would say at least x=2000 maybe more. Or maybe Cat 0 should be expanded a bit, and only Cat 0 restricted from TF/P-1.

I think at the end of the day we're trying to establish a policy such that:

1. Those who want to TF at the wavefront(s) (*in a timely manner*) are able to do so in a coordinated fashion.

1.1. Coordination helps minimize toe stepping.

1.2. Ideally this is automated. Never send a human to do a machine's job.


2. Those who reserve wavefront TF'ing work don't block actual FC efforts.

2.1. If a candidate isn't "optimally TF'ed" when it's given to an FC'er, it isn't the end of the world.


As always, GPU72 operates at the sufferance of George and this community. We will always "duck out" if we're slowing anything down.