TF Limits to Release At - Page 3

Prime95 · 2011-12-29, 01:30

Quote:

Originally Posted by Dubslow

Well if we can get the TF work through manual assignments, why would we need to go through GPU272? That reserves it as an LL, which is unnecessary due to your changes?

Edit: In other words, if GPU272 reserves them all, then the LL wavefront will be artificially pushed to 60M, where there's even less factoring done.

I'm not familiar with GPU272 inner workings. Why would it reserve more assignments than it needs? Why can't it reserve these as TF assignments rather than LL assignments?

Dubslow · 2011-12-29, 01:43

I think we need chalsall here. As far as I knew, the initial premise was to make it easier to get very low factoring work on expiries. As such, the system runs an LL collection spider (public name of GPU Factoring) just after 0000 UTC to collect expiries. TF work is then assigned via his system, and when complete, the LL is unreserved as another expiry, except now factored much higher. Example: http://www.mersenne.org/report_expon...&B1=Get+status Under the GPU272 system, I have this assigned for TF from 69 to 70 bits, to be complete in a day or two, after which it will be reassigned to someone else for TF to 72, and then P-1, before being released as an expiry.

so chalsall: Is this the system used for 58M? If so, how hard would it be to reserve them as TF instead of LL? Is that already being done?

Edit: @George: The Available Assignments and Overall System Progress might be of use. Note that there are currently 26K exponents reserved from PrimeNet, 8.5K of which are currently assigned to GPU272 workers, the system having already released around 9.5K as fully factored. Chalsall will have to answer as to whether or not exponents are released early.

davieddy · 2011-12-29, 06:43

A year or two ago, Oliver announced as "proof of concept" a program for GPUs which could TF quite fast. With similar modesty, Minigeek noted that this was ~100x faster than neutrinos.
I, (and presumably everyone else), thought "Hmm. Interesting. Seems like GPUs could usefully do a few more bits".

What could go wrong?

David

chalsall · 2011-12-29, 18:14

Quote:

Originally Posted by Dubslow

I think we need chalsall here.

Let me get back to you all later today. Family is arriving from The Great White North shortly, and we're doing a massive house cleaning....

davieddy · 2011-12-30, 05:31

Quote:

Originally Posted by chalsall

With all due respect, I don't understand why you are so fixated on "new" assignments vs. "recycled" (read: expired). At the end of the day, all of these have not had a LL test completed, and need to be worked.

If I state the "Bleedin' Obvious" one more time, Garo has assured me I will go to the chair.

And if you want to play God, GET YOUR ***KING ARSE IN GEAR.

David

garo · 2011-12-30, 10:05

The definition of insanity is to keep doing the same thing and expect a different result.

BigBrother · 2011-12-30, 10:11

Quote:

Originally Posted by davieddy

If I state the "Bleedin' Obvious" one more time, Garo has assured me I will go to the chair.

And if you want to play God, GET YOUR ***KING ARSE IN GEAR.

David

Your obnoxious posts have caused me to move all my resources to the GPU to 72 project.

Dubslow · 2011-12-30, 18:43

Quote:

Originally Posted by BigBrother

Your obnoxious posts have caused me to move all my resources to the GPU to 72 project.

Woo!

2011-12-29, 01:43	#24
Dubslow Basketry That Evening! "Bunslow the Bold" Jun 2011 40<A<43 -89<O<-88 1C35₁₆ Posts	I think we need chalsall here. As far as I knew, the initial premise was to make it easier to get very low factoring work on expiries. As such, the system runs an LL collection spider (public name of GPU Factoring) just after 0000 UTC to collect expiries. TF work is then assigned via his system, and when complete, the LL is unreserved as another expiry, except now factored much higher. Example: http://www.mersenne.org/report_expon...&B1=Get+status Under the GPU272 system, I have this assigned for TF from 69 to 70 bits, to be complete in a day or two, after which it will be reassigned to someone else for TF to 72, and then P-1, before being released as an expiry. so chalsall: Is this the system used for 58M? If so, how hard would it be to reserve them as TF instead of LL? Is that already being done? Edit: @George: The Available Assignments and Overall System Progress might be of use. Note that there are currently 26K exponents reserved from PrimeNet, 8.5K of which are currently assigned to GPU272 workers, the system having already released around 9.5K as fully factored. Chalsall will have to answer as to whether or not exponents are released early. Last fiddled with by Dubslow on 2011-12-29 at 01:47

2011-12-29, 06:43	#25
davieddy "Lucan" Dec 2006 England 2×3×13×83 Posts	It ain't rocket science A year or two ago, Oliver announced as "proof of concept" a program for GPUs which could TF quite fast. With similar modesty, Minigeek noted that this was ~100x faster than neutrinos. I, (and presumably everyone else), thought "Hmm. Interesting. Seems like GPUs could usefully do a few more bits". What could go wrong? David

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
P-1 limits?	nucleon	Hardware	8	2015-04-25 23:01
Musing on TF limits	davieddy	Lounge	10	2012-12-04 18:43
P-1 limits for Fermat numbers	Syd	Factoring	3	2011-05-23 11:13
GenefX64 limits	siegert81	Miscellaneous Math	2	2011-02-17 13:37
Proth Test Limits	amcfarlane	Math	1	2006-07-23 18:29

2011-12-30, 10:05	#28
garo Aug 2002 Termonfeckin, IE 5314₈ Posts	The definition of insanity is to keep doing the same thing and expect a different result.