mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > PrimeNet > GPU to 72

Reply
 
Thread Tools
Old 2018-05-24, 00:52   #4137
storm5510
Random Account
 
storm5510's Avatar
 
Aug 2009
U.S.A.

1,549 Posts
Default

Quote:
Originally Posted by chalsall View Post
...But if anyone has any resources they could allocate to LLTF'ing (preferably to at least 75 bits) it would be appreciated.

I can run 75's on this 1080. Give me an area, if the current 92 to 93 millions range is not what you need. I can do a 271 to 275 in 98 minutes, more or less. This is with the hardware in its factory configuration; roughly 1,050 GHz-d/day.


I had problems with Misfit and the GPUto72 batch process. I kept getting duplicate assignments. No issues with manual. I will just stick with that.



Last fiddled with by storm5510 on 2018-05-24 at 01:09 Reason: Correction.
storm5510 is online now   Reply With Quote
Old 2018-05-24, 01:22   #4138
chalsall
If I May
 
chalsall's Avatar
 
"Chris Halsall"
Sep 2002
Barbados

2×4,657 Posts
Default

Quote:
Originally Posted by storm5510 View Post
I can run 75's on this 1080. Give me an area, if the current 92 to 93 millions range is not what you need.
Coolness. 92M won't be needed for a couple of years; 87M is where the "crunch" is currently (because of the P-1'ers, who are nipping at our heels).

If you choose the "What Makes Sense" option to 75 bits, that would be great! Thanks.

Quote:
Originally Posted by storm5510 View Post
I had problems with Misfit and the GPUto72 batch process. I kept getting duplicate assignments.
Weird. As mentioned, GPU72 will never give out duplicate assignments with the exception of where a candidate is assigned and then unreserved. But if manual is working for you and you don't mind doing it, great!

P.S. Just so you know, the difference between "What Makes Sense" and "Let GPU72 Decide" is the former will never give work past the pledged TF level, while the latter sets the pledge level. Why? Historical design decision.
chalsall is offline   Reply With Quote
Old 2018-05-24, 02:10   #4139
kladner
 
kladner's Avatar
 
"Kieren"
Jul 2011
In My Own Galaxy!

23×32×139 Posts
Default

I have some queued for when the 74bit assignments run out on one of two workers (~320 GHzd/d each) on a GTX 1060. The 460 (220 Gd/d) has some lined up after the current assignment.

I did not pay attention, at first, and was given 75-76 work. I have run one of those, and am going to unload the rest when I'm sure of their status.
kladner is offline   Reply With Quote
Old 2018-05-24, 04:36   #4140
storm5510
Random Account
 
storm5510's Avatar
 
Aug 2009
U.S.A.

1,549 Posts
Default

Quote:
Originally Posted by chalsall View Post
...
If you choose the "What Makes Sense" option to 75 bits, that would be great! Thanks.
Done and done. It gave me assignments in the 87.7M range. 274 to 275. mfaktc tells me 58 minutes each.

I look at the work distribution map several times a week. I noticed a lot of work being done in this area. LL is close, in the 85's, and P-1 is being ran now in the 87's.


storm5510 is online now   Reply With Quote
Old 2018-06-15, 14:59   #4141
chalsall
If I May
 
chalsall's Avatar
 
"Chris Halsall"
Sep 2002
Barbados

2·4,657 Posts
Default We're under a crunch!

Hey all. Just a quick update, and a plea!

We've been slowly loosing ground to the Primenet P-1'ers. They've been gaining on us for a while now, and we're currently only about a month ahead of them.

This isn't the end of the world; we can release some candidates TF'ed to "only" 75 bits to feed them, and then bring those where a factor wasn't found to take to 76 bits before releasing them again for the LL'ers.

But, this is a bit of a maintenance pain for me, and slightly less optimal for Primenet's throughput. P-1'ing takes slightly less time the higher the candidate is TF'ed.

So, if anyone has any GPU compute they'd be willing to throw GPU72's way, it would be very much appreciated.

Relatedly, those who choose "What Makes Sense" or "Let GPU72 Decide", the system will sometimes assign work to only 75 bits instead of the nominal 76, depending on the current demand vs. availability situation.
chalsall is offline   Reply With Quote
Old 2018-06-15, 17:06   #4142
ixfd64
Bemusing Prompter
 
ixfd64's Avatar
 
"Danny"
Dec 2002
California

28×32 Posts
Default

Quote:
Originally Posted by chalsall View Post
Hey all. Just a quick update, and a plea!

We've been slowly loosing ground to the Primenet P-1'ers. They've been gaining on us for a while now, and we're currently only about a month ahead of them.

This isn't the end of the world; we can release some candidates TF'ed to "only" 75 bits to feed them, and then bring those where a factor wasn't found to take to 76 bits before releasing them again for the LL'ers.

But, this is a bit of a maintenance pain for me, and slightly less optimal for Primenet's throughput. P-1'ing takes slightly less time the higher the candidate is TF'ed.

So, if anyone has any GPU compute they'd be willing to throw GPU72's way, it would be very much appreciated.

Relatedly, those who choose "What Makes Sense" or "Let GPU72 Decide", the system will sometimes assign work to only 75 bits instead of the nominal 76, depending on the current demand vs. availability situation.
I noticed a lot of people are still trial factoring DC candidates. Perhaps you could put a message on the DC TF assignment request page?

Last fiddled with by ixfd64 on 2018-06-15 at 17:06
ixfd64 is online now   Reply With Quote
Old 2018-06-15, 17:45   #4143
chalsall
If I May
 
chalsall's Avatar
 
"Chris Halsall"
Sep 2002
Barbados

100100011000102 Posts
Default

Quote:
Originally Posted by ixfd64 View Post
I noticed a lot of people are still trial factoring DC candidates. Perhaps you could put a message on the DC TF assignment request page?
Yeah, that's a good idea. I'll do that.

ktony has already moved most of his GPUs to LLTF'ing; I'm moving my three little GPUs up there later tonight.

There are some who specifically want to do DCTF'ing. And that's fine; it's their kit.
chalsall is offline   Reply With Quote
Old 2018-06-15, 17:47   #4144
Mark Rose
 
Mark Rose's Avatar
 
"/X\(‘-‘)/X\"
Jan 2013

2·1,433 Posts
Default

I've moved mine.
Mark Rose is offline   Reply With Quote
Old 2018-06-16, 01:30   #4145
Chuck
 
Chuck's Avatar
 
May 2011
Orange Park, FL

5·173 Posts
Default

Quote:
Hey all. Just a quick update, and a plea! We've been slowly loosing ground to the Primenet P-1'ers. They've been gaining on us for a while now, and we're currently only about a month ahead of them.
I dropped out of GPUto72 in February 2016 when Squirrels was submitting so much work that my contributions didn't seem very worthwhile. I had been TFing on James' >1000M site.

Since then my eight-year-old i7-970 machine went belly up after eight years (a good run) and I have since replaced it with an i7-7800X and a GTX 1080ti.

I had the old MISFIT structure still intact, so I just had to update some directory paths and copy in the the CUDA 8.0 mfaktc and dlls. It's running now. Getting about 1500 GHz-d/day from 74 to 76.
Chuck is offline   Reply With Quote
Old 2018-06-16, 02:07   #4146
Chuck
 
Chuck's Avatar
 
May 2011
Orange Park, FL

5·173 Posts
Default

Quote:
Originally Posted by Chuck View Post
I had the old MISFIT structure still intact, so I just had to update some directory paths and copy in the the CUDA 8.0 mfaktc and dlls. It's running now. Getting about 1500 GHz-d/day from 74 to 76.
OOPS! Just realized I was running the less-classes mfaktc version which is appropriate for the short runs on James' site. As soon as the first exponent finishes processing I will stop and copy in the correct version.
Chuck is offline   Reply With Quote
Old 2018-06-16, 03:18   #4147
chalsall
If I May
 
chalsall's Avatar
 
"Chris Halsall"
Sep 2002
Barbados

2×4,657 Posts
Default

Quote:
Originally Posted by Chuck View Post
I had the old MISFIT structure still intact, so I just had to update some directory paths and copy in the the CUDA 8.0 mfaktc and dlls. It's running now. Getting about 1500 GHz-d/day from 74 to 76.
Super coolness! Welcome back Chuck!
chalsall is offline   Reply With Quote
Reply

Thread Tools


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

All times are UTC. The time now is 16:28.

Tue Sep 29 16:28:34 UTC 2020 up 19 days, 13:39, 0 users, load averages: 1.79, 1.56, 1.59

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.