GPU to 72 status... - Page 386

chalsall · 2019-01-01, 02:24

Quote:

Originally Posted by James Heinrich

Hey, I just spent 6 months doing CPU-TF!

Of course, that was because no GPU software supports >2³² exponents...

LOL...

Whatever rocks your boat!

Uncwilly · 2019-01-01, 05:43

Quote:

Originally Posted by chalsall

Dude! You're still using a CPU to TF? Are you stoned, man?

My kit, my joy. Actually some of the machines that I am using I don't want to discover a prime on. TF ensures that. And on my laptop, it keeps the fan quiet and the machine cool. SDC and TC reduce the risk of finding a prime low enough that I am willing to risk it on those other machines.

chalsall · 2019-01-01, 13:35

Quote:

Originally Posted by Uncwilly

My kit, my joy.

Absolutely. Sorry, should have put a

or a

on my post....

kriesel · 2019-01-01, 14:27

Please consider running scattered islands of exponents up to or toward the full GPUto72 individual goal bit levels. You could call it something like QAislands.
Something like for any million range in exponent, the first thousand range of exponents.
For example, for the range 171 million to 172 million, take each prime exponent between 171000000 to 171001000 to the lesser of 78 bits or whatever finds a factor.

Users like mikr and rudimeier have already done some of this. It is very useful when prequalifying a few exponents for P-1 software testing on high exponents on gpu by finishing them myself to gputo72 factoring goal levels. Thank you to the pioneers who have already done some of this, to or near primenet goal bit levels several years ago, for example.
(Also, some of the computerids are entertaining. Bonnie, clyde, bart, homer, ...)

The higher ones will represent a very considerable amount of total work per exponent.
https://www.mersenne.ca/exponent/101000117 114GhzD to go to gputo72 level (76)
https://www.mersenne.ca/exponent/171000043 346GhzD to go to gputo72 level (78)
https://www.mersenne.ca/exponent/371000039 1.2ThzD to go to gputo72 level (81)
https://www.mersenne.ca/exponent/919000001 8.4ThzD to go to gputo72 level (85)
https://www.mersenne.ca/exponent/999999043 13.7 ThzD to go to gputo72 level (86)

chalsall · 2019-01-01, 15:59

Quote:

Originally Posted by kriesel

Please consider running scattered islands of exponents up to or toward the full GPUto72 individual goal bit levels. You could call it something like QAislands.

I understand the rationale for doing this kind of thing, but I'd prefer that GPU72 (mostly) focuses on the Primenet wave-fronts.

What I'd suggest you do is solicit assistance from forum participants by way of a thread or sub-forum. Similar to what Aaron does with Strategic Double Checks and Triple Checks.

kriesel · 2019-01-01, 17:12

Quote:

Originally Posted by chalsall

I'd prefer that GPU72 (mostly) focuses on the Primenet wave-fronts.

I agree. Emphatically. Keeping a substantial stock of adequately TFed exponents in front of the cpu wave is a top priority.

chalsall · 2019-01-07, 21:39

Quote:

Originally Posted by chalsall

We're about two weeks away from taking everything below 97M to at least 72 "bits". ... So, once everything "owned" by GPU72 at 71 is assigned, the system will start giving out work to 73 bits (starting in 92M), even if the pledge is only to 72.

OK, this was a little faster than expected, but GPU72 is now "out of" candidates to take to 72.

From now on, all automated requests (MISFIT, mloop et al) will automatically bump the pledge level up to 73.

For those who use the manual assignment page at GPU72, if they choose a pledge of 72 it will show "Sorry... No assignments available which match your criteria." in the preview. If they still click "Get Assignments" they will be assigned work to 73 bits (currently in 92M).

I'll monitor the logs for the next couple of hours, to make sure I haven't made a SPE. But, as always, please let me know if anyone sees anything strange....

petrw1 · 2019-01-07, 21:47

Quote:

Originally Posted by chalsall

OK, this was a little faster than expected, but GPU72 is now "out of" candidates to take to 72.

From now on, all automated requests (MISFIT, mloop et al) will automatically bump the pledge level up to 73.

For those who use the manual assignment page at GPU72, if they choose a pledge of 72 it will show "Sorry... No assignments available which match your criteria." in the preview. If they still click "Get Assignments" they will be assigned work to 73 bits (currently in 92M).

I'll monitor the logs for the next couple of hours, to make sure I haven't made a SPE. But, as always, please let me know if anyone sees anything strange....

Just curious: why stop at 97M....vs for example 100M ?

Mark Rose · 2019-01-07, 21:54

Might be worth it to go up to <101M.

chalsall · 2019-01-07, 22:25

Quote:

Originally Posted by Mark Rose

Might be worth it to go up to <101M.

As explained above, some are working "off the books" in 97M and above, and I always get yelled at (wink) when GPU72 assigns work which is then completed by someone else....

LaurV · 2019-01-09, 04:09

Quote:

Originally Posted by chalsall

As explained above, some are working "off the books" in 97M and above, and I always get yelled at (wink) when GPU72 assigns work which is then completed by someone else....

This will always be the case, no matter what you do, somebody will yell at you. Ask those who complain to raise their pledge, to avoid "of the book" work (mainly done by people who are after fast factors, so easy tasks, low bitlevels).

2019-01-01, 14:27	#4239
kriesel "TF79LL86GIMPS96gpu17" Mar 2017 US midwest 152B₁₆ Posts	Scattered full TF depth Please consider running scattered islands of exponents up to or toward the full GPUto72 individual goal bit levels. You could call it something like QAislands. Something like for any million range in exponent, the first thousand range of exponents. For example, for the range 171 million to 172 million, take each prime exponent between 171000000 to 171001000 to the lesser of 78 bits or whatever finds a factor. Users like mikr and rudimeier have already done some of this. It is very useful when prequalifying a few exponents for P-1 software testing on high exponents on gpu by finishing them myself to gputo72 factoring goal levels. Thank you to the pioneers who have already done some of this, to or near primenet goal bit levels several years ago, for example. (Also, some of the computerids are entertaining. Bonnie, clyde, bart, homer, ...) The higher ones will represent a very considerable amount of total work per exponent. https://www.mersenne.ca/exponent/101000117 114GhzD to go to gputo72 level (76) https://www.mersenne.ca/exponent/171000043 346GhzD to go to gputo72 level (78) https://www.mersenne.ca/exponent/371000039 1.2ThzD to go to gputo72 level (81) https://www.mersenne.ca/exponent/919000001 8.4ThzD to go to gputo72 level (85) https://www.mersenne.ca/exponent/999999043 13.7 ThzD to go to gputo72 level (86) Last fiddled with by kriesel on 2019-01-01 at 14:32

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

2019-01-07, 21:54	#4244
Mark Rose "/X\(‘-‘)/X\" Jan 2013 2·5·293 Posts	Might be worth it to go up to <101M.