Davieddy's World

[davieddy]

Originally Posted by chalsall

There are 206 candidates in the 62M range which are "pledged" to only 72. Plus another 99 between 60M and 63M TFed to less than 73 which we might get access to (they're currently assigned by Primenet).

But, yes, overall we're effectively done (or, at least, fully assigned) below 63M.

LL assignment rate needs to to be at least as fast as LL completion rate (Confucious).
I know you described this as a non-sequitur a while back, but then you are from Barbados.
LL assignment is in such a mess, the average rate is tricky to monitor.
Just poach all those lagging, hoarded "one bit at a time" Primenet TF assignments.
This is what I meant when I asked you and George to get your act together.

How do you intend to TF to 74 at the required rate (1000 per 3 days ATM?).

I won't pull any more punches until you attempt a thoughtful answer to these points.

David

[Prime95]

Originally Posted by davieddy

Just poach all those lagging, hoarded "one bit at a time" Primenet TF assignments.

Is this a problem? I don't have time to quantify this right now, but I remember eyeballing a subset of the current assignments a few months ago and I didn't think there were too many lagging TFs assigned to prime95 clients.

[chalsall]

Originally Posted by davieddy

How do you intend to TF to 74 at the required rate (1000 per 3 days ATM?).

Watch us. We can just do it.

Estimated Completion of 54,475 candidates: 160.07 days. Primenet LL completion time (accepting your ~333 a day): 163.59 days.

Two things not taken into account is, first, we'll also eliminate a few thousand candidates within this block, so the number needed to be LLed will be less (works against our staying ahead). Second, LLing takes longer the larger the candidate (works in favor of our staying ahead).

And, of course, we can always return a few hundred TFed to "only" 73 if needed.

Edit: Two additional things to consider... There are appoximately 10,600 candidates already appropriately TFed held by Primenet for assignment below 65M (as in, we're currently ~32 days ahead). Plus, approximately 16,000 candidates are expected to expire in the next eleven days.

[chalsall]

Originally Posted by Prime95

Is this a problem? I don't have time to quantify this right now, but I remember eyeballing a subset of the current assignments a few months ago and I didn't think there were too many lagging TFs assigned to prime95 clients.

Not really.

But there are a few which have been assigned for a ridiculous amount of time, with no apparent progress. (And, everything below 58M has already been "poached", with both TFing and P-1'ing completed.)

[davieddy]

Originally Posted by chalsall

Estimated Completion of 54,475 candidates: 160.07 days. Primenet LL completion time (accepting your ~333 a day): 163.59 days.

In the last week:
2147 LL tests completed.
1173 TFed to 73.
922 TFed to 74.
(Your data).

On what is your estimate based?

D

[chalsall]

Originally Posted by davieddy

In the last week:
2147 LL tests completed.
1173 TFed to 73.
922 TFed to 74.
(Your data).

On what is your estimate based?

D

OK David. Let's "cut to the chase".

What would you advise we do that we're not already doing?

[Aramis Wyler]

Originally Posted by davieddy

In the last week:
2147 LL tests completed.
1173 TFed to 73.
922 TFed to 74.

I don't see much problem here, myself. Even if we only did 2147/week (306/day) and even if the masses were able to maintain the 333/day completion rate you gave despite the numbers getting bigger, our estimated completion of the 54475 candidates is still only 177.6 days vs the stated 160 days. Since we're 32 days ahead now, that means that after 6 months of factoring to 74 we'll only be 18 days ahead. Boo f'ing hoo. If in 6 months our hardware hasn't sped up relative to the croud, we may have to spend a few weeks doing to 73. What have we lost in that case? Nothing but some surplus lead time. And what have we gained? 6 months of doing a better job factoring than we were last week. Sounds like a pretty good case to me.

[davieddy]

Originally Posted by Aramis Wyler

I don't see much problem here, myself. Even if we only did 2147/week (306/day) and even if the masses were able to maintain the 333/day completion rate you gave despite the numbers getting bigger, our estimated completion of the 54475 candidates is still only 177.6 days vs the stated 160 days.

The lunatics have taken over the asylum.
306 LLs/day is what us great unwashed are completing currently. Average exponent ~57M.
333 TFs/day is what ye almighty should be doing (Chalsall misread as usual). Exponent 63M.
What you are doing is 299TFs/day, the majority being to 73 bits.

In a year's time, we will be completing 336 LLs per day or better, average exponent ~63M.
You should be aiming for 366 TFs per day around 70M.

THX largely to AVX, these are good times for GIMPS.
Get with the action man, and forget 74 bits as the default for now.

David

[chalsall]

Originally Posted by davieddy

Get with the action man, and forget 74 bits as the default for now.

Your above didn't actually answer any questions.

What would you advise we do?

Why?

[kladner]

Originally Posted by chalsall

There are 206 candidates in the 62M range which are "pledged" to only 72. Plus another 99 between 60M and 63M TFed to less than 73 which we might get access to (they're currently assigned by Primenet).

But, yes, overall we're effectively done (or, at least, fully assigned) below 63M.

The ones I have are pledged to 73 as the tables indicated they should be.

[davieddy]

Originally Posted by chalsall

Your above didn't actually answer any questions.

What would you advise we do?

Why?

Can you read?

Answer mine first.
What are your estimated completion times based on?

I advise you stick to 73 bits until you are comfortably clear of the confused LL assignment front, then do some to 74 as resources allow.

x