The "(W)TF Depth Debate" state of the bunion - Page 15

davieddy · 2013-12-20, 06:04

Quote:

Originally Posted by kladner

But irony in what direction, and to what end?

Irony for irony's sake, since Chris hasn't dared to engage in serious discussion.
With the notable exception of Aramis, Foetus Boy seems more up to speed on this topic than anyone except me, and it seems that includes George.

You will understand this mildy mathy point though:

From the point of view of the LL assignment seeker, a 69M exponent TFed to 74 will take 30% longer to test than a 60M exponent TFed to 73.
Furthermore it will be (60/69)*(74/73) = 0.88 times as likely to be prime.

That is what I meant by "unappetizing fare", and this cannot do anything but slow the assignment rate.
Remember that maximizing this rate is the sole purpose of these heady bit levels and of this debate.

D

kladner · 2013-12-20, 12:00

Quote:

Originally Posted by Batalov

The other way around.
Mr.Show (1995–1998)
Sopranos (1999-2007) is Mr.Show without the laugh track.

Point taken. Wow! I had no idea that The Sopranos ran that long.

davieddy · 2013-12-20, 12:43

Quote:

Originally Posted by Prime95

And one final note. Never give out a TF assignment for a bit depth greater than recommended by James' TF vs. LL break-even chart.

What James' stepped curve shows is that GPUs should be directed entirely to TF, and that the bit level should be determined by available firepower.
However James' steps occur when the exponent increases by 2^1/3, and so do David's firepower-limited steps.
Now ain't that as cute as it is unsurprising?

Quote:

Thus, with unlimited GPU firepower we'd factor everything up to James' limits. With less than unlimited firepower, we'd ~~always do the TF with the biggest payoff~~ factor everything up to David's limits.

Heck, even davieddy might agree with such a strategy.

With the proviso that

Quote:

In practice, will implementing such a strategy make much of an impact? Are there significant numbers of exponents in the 55M-64M range factored to only 2^72 being released by PrimeNet each night?

No. But if you and Chris wish to persist ~~tidying up~~ mucking about behind the wavefront, the cost is raising my cut-off points.

Heck, even garo might agree with such a strategy.

davieddy · 2013-12-20, 15:17

Some people might like a formula for James' and my curves.

It is:
bitlevel is proportional to (TF firepower/LL firepower) * exponent³ .

END OF STORY.

BTW do you think there are any GPUs out there which don't belong to GPUto72?

davieddy · 2013-12-20, 15:34

Quote:

Originally Posted by davieddy

Whoopee!

In response to the daily threat of running out of available candidates, Chris has released 300 60M and 61M candidates which (PLEASE GOD) will have been TFed only to 73.

His usual strategy is to slow down the assignment rate by bagging everything up to 68M, leaving only unappetizing fare during the peak demand.

What a ****.

Sure enough 1500 available LL assignments between 63M and 69M went AWOL between 1400 and 1500 just now.

The crowd outside are clamouring for bread, Marie Antoinette.
"Let them eat soixante-neuf M exponents."

chalsall · 2013-12-20, 16:49

Quote:

Originally Posted by davieddy

I sense that one of us is resorting to bullying in indesperate[sic] fear of being exposed as a fraud.

I concur.

chalsall · 2013-12-20, 17:01

Quote:

Originally Posted by davieddy

The crowd outside are clamouring for bread, Marie Antoinette.
"Let them eat soixante-neuf M exponents."

1. Everything being assigned for LL'ing has been TFed to at least 74, and P-1ed.

2. This work (read: LLing) needs to be done.

3. Over 50% of LL assignments won't get a single cycle done on it.

3.1. Over 80% of LL assignments will be "recycled" after it is abandoned.

4. Where, approximately (based on some hard math, please), is the next MP expected to be? In the 60s, or 70s?

5. We have the fire-power to pull this (read: TFing to 74) off!

davieddy · 2013-12-20, 18:12

Quote:

Originally Posted by chalsall

1. Everything being assigned for LL'ing has been TFed to at least 74, and P-1ed.

2. This work (read: LLing) needs to be done.

3. Over 50% of LL assignments won't get a single cycle done on it.

3.1. Over 80% of LL assignments will be "recycled" after it is abandoned.

4. Where, approximately (based on some hard math, please), is the next MP expected to be? In the 60s, or 70s?

5. We have the fire-power to pull this (read: TFing to 74) off!

1. Pigs still airborne (just).
2. and LLers like to do it in ascending order.
3. No because they want a LOWER one.
4. Poisson. ~50% chance of it being <80M.
5. Read this thread you created.

SEND HER TO THE GUILLOTINE

owftheevil · 2013-12-20, 20:15

Actually, some of us want to do it in descending order, but we can't figure out how to count backwards from $\omega$ .

kladner · 2013-12-20, 20:46

Quote:

Originally Posted by owftheevil

Actually, some of us want to do it in descending order, but we can't figure out how to count backwards from $\omega$ .

chalsall · 2013-12-20, 22:43

Quote:

Originally Posted by davieddy

SEND HER TO THE GUILLOTINE

Please define "HER".

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Stockfish game: "Move 8 poll", not "move 3.14159 discussion"	MooMoo2	Other Chess Games	5	2016-10-22 01:55
Aouessare-El Haddouchi-Essaaidi "test": "if Mp has no factor, it is prime!"	wildrabbitt	Miscellaneous Math	11	2015-03-06 08:17
Specifing TF factor depth in "Manual Assignments"?	kracker	PrimeNet	2	2012-07-22 17:49
Would Minimizing "iterations between results file" may reveal "is not prime" earlier?	nitai1999	Software	7	2004-08-26 18:12

2013-12-20, 15:17	#158
davieddy "Lucan" Dec 2006 England 2·3·13·83 Posts	Some people might like a formula for James' and my curves. It is: bitlevel is proportional to (TF firepower/LL firepower) * exponent³ . END OF STORY. BTW do you think there are any GPUs out there which don't belong to GPUto72?

2013-12-20, 20:15	#163
owftheevil "Carl Darby" Oct 2012 Spring Mountains, Nevada 315₁₀ Posts	Actually, some of us want to do it in descending order, but we can't figure out how to count backwards from $\omega$ .