| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
| World Community Grid Forums
|
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 53
|
|
| Author |
|
|
ryan222h
Senior Cruncher Joined: Sep 4, 2006 Post Count: 425 Status: Offline |
Jim, it must be something with your system.
----------------------------------------I have multiple 7770's and they are running 4 wu at a time, no errors or invalids.... ![]() |
||
|
|
Jim1348
Veteran Cruncher USA Joined: Jul 13, 2009 Post Count: 1066 Status: Offline Project Badges:
|
Yes, it could probably be solved with enough sacrifice of other projects, but it is doing what it is supposed to do on my system. It has also been suggested that over-volting the HD 7770 might help when running multiple work units, but it would take a certain amount of experimenting to find out what fixes it. I could only run two anyway with a single CPU (virtual) core, and it is not worth it. The whole project will be over before it is fixed.
----------------------------------------[Edit 1 times, last edit by Jim1348 at Nov 12, 2012 4:07:11 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
@Jim,
If you want to do other projects with those AMD cards, Just run 1 HCC1 WU each. I have not been able to get any other app_info working that will get me anything but HCC1 consistently. However, if you wish to direct all efforts towards HCC1, put your ATI/AMD cards in one machine and let them fly with multitasking parrallel processing. Points be damned, it's the most efficient way to get the science returned in the fastest manner :) |
||
|
|
Jim1348
Veteran Cruncher USA Joined: Jul 13, 2009 Post Count: 1066 Status: Offline Project Badges:
|
shdbcamping,
Yes, those are the choices. I am doing the single work unit option on two cards. But there is no reason for me to sacrifice other projects for this one; they get back faster too if I devote resources to them. And no one knows which is more worthwhile in the end. I am wondering whether the HCC1 scientists are even prepared for the deluge they are getting. Can they make use of it coming back at once so soon? It will be interesting to see if they do a follow-up, but the new HD 8000 series will be out around that time anyway and the tradeoffs will be a little different. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
See today's post by Viktors on remaining work estimate. Do not worry about the scientists and the now ~6 times faster coming returns. They've been working on GPUfying HCC for probably 2 years now.
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
But as Jim1348 stated in his question, I'm only running one 7870 with 12 instances... and I have in excess of 54 pages of "Pending Verification". What will it be when my 7950 comes online? 110 pages? how are we moving this fast forward with this amount of unverified results? I have to ask because I have a BS in biology and do not understand how targets could be confirmed with so many millions of previous results unverified ;(
|
||
|
|
mmstick
Senior Cruncher Joined: Aug 19, 2010 Post Count: 151 Status: Offline Project Badges:
|
shbdcamping, make that 200 pages of pending validation, that is what you will see if you get the 7950 working in optimal condition alongside your 7870. You may even want to try doubling to 24 instances to solve any memory inconsistencies. I currently run 16 instances for my single 7950 on my FX-8120.
----------------------------------------Everything pending validation/verification is merely waiting for your wingmen to catch up with you. Verification means one or two of your wingmen goofed with errors/invalids. When you have this level of computing power, all those people still doing CPU crunching will weigh you down, as they can take hours to days to complete a specific work unit, while I know I eat one work unit every 22 seconds. In due time, your results will eventually validate/verify. In the meantime, try to make a world record for number of work units pending at a given time. [Edit 1 times, last edit by mmstick at Nov 12, 2012 10:05:09 PM] |
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges:
|
As a quick update, we have just launched a quick batch of 2 images per workunit out into the wild as a quick test. There is about 18 hours more of 1 image per workunit jobs left and then all new jobs will contain 2 images per workunit.
|
||
|
|
Bearcat
Master Cruncher USA Joined: Jan 6, 2007 Post Count: 2803 Status: Offline Project Badges:
|
I currently run 16 instances for my single 7950 on my FX-8120. Am curious how high your fans need to be running 16. Am currently running 10 on both 7950 and 7870 and let the card run the fans as needed. Does a better job than my 6950 I had. As for PV jail, am currently running above 100 pages. BTW, whats the setting to crunch 12? Might let the 7950 try it.
Crunching for humanity since 2007!
----------------------------------------![]() [Edit 1 times, last edit by Bearcat at Nov 13, 2012 12:04:58 AM] |
||
|
|
mmstick
Senior Cruncher Joined: Aug 19, 2010 Post Count: 151 Status: Offline Project Badges:
|
I currently run 16 instances for my single 7950 on my FX-8120. Am curious how high your fans need to be running 16. Am currently running 10 on both 7950 and 7870 and let the card run the fans as needed. Does a better job than my 6950 I had. As for PV jail, am currently running above 100 pages. As an enthusiast hobbyist and overclocker, I run no less than the best cooling solutions at 100% speed at all times with minimal noise level due to efficient fans and heatsinks for CPU and GPU. Max temperature of CPU is 50C Max temperature of GPU running 100% fan is 50C Max temperature of GPU running 50% fan is 68C. GPUs will crunch fine even in 100C, although the higher your overclock, the lower your threshold of error is with temperature. My 7950 is one of those Gigabyte Windforce 3's, so it has three low noise fans that provide a rather high CFM together. I was a bitcoin miner before I was here, so I was running this monster at 100% load 24/7 since March. 1/12 = GPU count of 0.083 [Edit 1 times, last edit by mmstick at Nov 13, 2012 12:29:16 AM] |
||
|
|
|