GPU to 72 status... - Page 371

Mark Rose · 2017-09-13, 22:04

GPU72 isn't picking up that I've completed my assignment for 85137103. I had submitted the bit levels out of orders due to reasons, and it appears GPU72 is choking on that.

chalsall · 2017-09-14, 00:05

Quote:

Originally Posted by Mark Rose

GPU72 isn't picking up that I've completed my assignment for 85137103.

Actually, Primenet is reporting that candidate only TF'ed to 75. If you resubmit your 75 to 76 work manually, Primenet and GPU72 should both be happy.

James Heinrich · 2017-09-14, 00:23

It's not so much a bug in PrimeNet as it not being as smart as perhaps it could be.
The exponent was previously taken to 74, then Mark submitted 75-76 and PrimeNet said "that's nice", accepted the result for the discontinuous bit tested but keeps the how-far-TF'd value at 74 since 74-75 has not been reported done. Later Mark says 74-75 has been done so PrimeNet updates how-far-TF'd from 74->75 but doesn't go back to check if some time in the past any higher bitlevels have been done out of order.

As Chris suggested, re-submitting 75-76 should fix the problem.

Mark Rose · 2017-09-14, 02:45

Done.

Got more credit on PrimeNet, too.

Prime95 · 2017-09-14, 19:43

Does GPU72 have any of 46M to 50M checked out for TF? The Primenet status map shows 33,000 exponents reserved for TF.

petrw1 · 2017-09-14, 23:00

Short answer: Yes.

I happened to mention a few weeks ago that I was going to TF a little in the 48/9M range what was below the "YELLOW" line; Chris offered to open this up on GPU72 as it would help DC.

It seems he has since extended that back to 46M for non-DC'd exponents.

That is all I know...

If you are suggesting he swallowed up too many and left none for DC you'll need to take that up with him.

Prime95 · 2017-09-14, 23:06

Quote:

Originally Posted by petrw1

If you are suggesting he swallowed up too many and left none for DC you'll need to take that up with him.

I do not know if that is the case. I got an email asking what was going on and I was surprised to see the reserved numbers as GPU72 finished the DC range a long time ago.

Mark Rose · 2017-09-15, 00:39

Quote:

Originally Posted by Prime95

I do not know if that is the case. I got an email asking what was going on and I was surprised to see the reserved numbers as GPU72 finished the DC range a long time ago.

It's from two things:

We have too much TF throughput relative to LL
Newer cards are even more skewed towards TF over LL, changing the crossover point

so might as well do a little more TF before the DC.

linament · 2017-09-15, 16:31

Is GPU72 supposed to be handing out DC candidates that were already verified? For example, 48860351 and 48856421.

chalsall · 2017-09-15, 17:23

Quote:

Originally Posted by petrw1

It seems he has since extended that back to 46M for non-DC'd exponents.

Yes. Looking at James' GPU cross-over analysis, it makes sense to TF from 45M and up to 73 bits on most cards. Since we are so far ahead of LLTF'ing, and some are interested in doing more DCTF'ing, I've brought in available candidates for this work.

Please keep in mind that almost all DC "Cat 4" candidates (currently above 46123352) are assigned to the "churners", who almost never complete their assignments.

chalsall · 2017-09-15, 17:25

Quote:

Originally Posted by linament

Is GPU72 supposed to be handing out DC candidates that were already verified?

Hmmm... No, that's weird. Let me drill down on that.

2017-09-15, 16:31	#4079
linament Nov 2013 2·3² Posts	Verified Candidates Is GPU72 supposed to be handing out DC candidates that were already verified? For example, 48860351 and 48856421.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Status	Primeinator	Operation Billion Digits	5	2011-12-06 02:35
62 bit status	1997rj7	Lone Mersenne Hunters	27	2008-09-29 13:52
OBD Status	Uncwilly	Operation Billion Digits	22	2005-10-25 14:05
1-2M LLR status	paulunderwood	3*2^n-1 Search	2	2005-03-13 17:03
Status of 26.0M - 26.5M	1997rj7	Lone Mersenne Hunters	25	2004-06-18 16:46

2017-09-13, 22:04	#4071
Mark Rose "/X\(‘-‘)/X\" Jan 2013 101101110001₂ Posts	GPU72 isn't picking up that I've completed my assignment for 85137103. I had submitted the bit levels out of orders due to reasons, and it appears GPU72 is choking on that.

2017-09-14, 00:23	#4073
James Heinrich "James Heinrich" May 2004 ex-Northern Ontario CAB₁₆ Posts	It's not so much a bug in PrimeNet as it not being as smart as perhaps it could be. The exponent was previously taken to 74, then Mark submitted 75-76 and PrimeNet said "that's nice", accepted the result for the discontinuous bit tested but keeps the how-far-TF'd value at 74 since 74-75 has not been reported done. Later Mark says 74-75 has been done so PrimeNet updates how-far-TF'd from 74->75 but doesn't go back to check if some time in the past any higher bitlevels have been done out of order. As Chris suggested, re-submitting 75-76 should fix the problem.

2017-09-14, 02:45	#4074
Mark Rose "/X\(‘-‘)/X\" Jan 2013 29×101 Posts	Done. Got more credit on PrimeNet, too.

2017-09-14, 19:43	#4075
Prime95 P90 years forever! Aug 2002 Yeehaw, FL 7²·149 Posts	Does GPU72 have any of 46M to 50M checked out for TF? The Primenet status map shows 33,000 exponents reserved for TF.

2017-09-14, 23:00	#4076
petrw1 1976 Toyota Corona years forever! "Wayne" Nov 2006 Saskatchewan, Canada 2³×3⁴×7 Posts	Short answer: Yes. I happened to mention a few weeks ago that I was going to TF a little in the 48/9M range what was below the "YELLOW" line; Chris offered to open this up on GPU72 as it would help DC. It seems he has since extended that back to 46M for non-DC'd exponents. That is all I know... If you are suggesting he swallowed up too many and left none for DC you'll need to take that up with him.