BOINC

I can crunch at GPUGRID with my 750Ti cards now? You showed short runs. Do they have long run work for the 750Ti cards as well?
 
AgrFan had an issue back around those dates with his account getting hijacked. I think he got it resolved, but only he could verify that. http://hardforum.com/showpost.php?p=1040690279&postcount=1650

As far as your stats go, WCG points translate into BOINC points * 7.


Short runs only last I read. The long runs still use CUDA 5.5 which error out. You will need to disable long runs for it to work.
 
Last edited:
I remember AgrFan being hacked, I was just unsure because it doesn't show on that page his movement back to [H]ardOCP.

Because nothing I have tried will get me more than 4 GPUGRID WUs and I only allow transfers during the night, I found that long WUs are a better choice for me. Short runs will leave my cards doing nothing most of the day unless I use Einstein as backup.
 
WCG forums have it discussed multiple times. What happened is when they switched from UD to BOINC years ago, the points didn't translate well to the then "current" point structure. So, WCG has had it on their plate to "fix" the disparity but it is like most user requested changes. It just doesn't happen. Or I should say happen in a very timely manner. But it is pretty consistent that 7 WCG points = 1 BOINC point.

Other projects do not do that because they started their points on BOINC. I am still waiting to hear back from theSkyNet about how they plan on doing their points once they move their Source Finder app to BOINC. This info was not in the email they sent me. Therefore, it could potentially have this issue or could just be thrown in as is... Only time will tell.


Out WCG team page at WCG still shows AgrFan as a team member. So, I'm sure he is paying attention.

I'm also guessing they are limiting the number of work units at GPUGrid since they want results returned quickly and don't want people with slow cards sitting on large caches.
 
Gilthanis said:
AgrFan had an issue back around those dates with his account getting hijacked. I think he got it resolved, but only he could verify that. http://hardforum.com/showpost.php?p=1040690279&postcount=1650
Click to expand...

I'm definitely crunching for HardOCP.
http://www.hardfolding.com/index.php?p=2611&id=43058&tid=1411

Changing my password resolved the issue. WCG found the hacker a short time later.
https://secure.worldcommunitygrid.org/forums/wcg/viewthread_thread,36475_offset,20#451377
 
Gilthanis said:
It appears that Ripple Labs will be discontinuing their Computing for Good Giveaways. Thus will be no longer paying its team their cryptocurrency called XRP.

https://ripple.com/forum/viewtopic.php?f=18&t=6343

http://www.worldcommunitygrid.org/forums/wcg/viewthread_thread,36557
Click to expand...
If you're looking to get rewarded for boinc computations you should look towards gridcoin, a cryptocurrency that rewards users for their boinc contributions.

www.reddit.com/r/gridcoin
http://gridcoinnetwork.org/
https://cryptocointalk.com/forum/464-gridcoin-grc/
 
Beal@home now has a Windows app. I believe these work units require 2GB of RAM
 
Rosetta results posted.

I don't know if it's been posted before, but over at Rosetta, they have one of their scientific and technical writers posting monthly updates surrounding Rosetta@home's contributions to the Baker Labs research.

Click Here for that magical thread.

Also, 28 days ago the lead researcher over at LeidenClassical posted about whats being researched at the moment.
 
Last edited:
Kai said:
Rosetta results posted.

I don't know if it's been posted before, but over at Rosetta, they have one of their scientific and technical writers posting monthly updates surrounding Rosetta@home's contributions to the Baker Labs research.

Click Here for that magical thread.

Also, 28 days ago the lead researcher over at Oproject posted about whats being researched at the moment.
Click to expand...

I think you meant Leiden rather than OProject...
 
Yeah... I did that on purpose, to test you guys!!!!!!!!!









Doh!
 
Yeah...the project admin at OProject (and now also Beal@home) doesn't impress me too much. He stopped responding to posts at OProject, has let that project go to crap, presented an idea for a horrible BOINC point stock exchange, and now is starting another project that he could have just added as a sub project to the one he already has. When I asked him at BOINCStats why he didn't just make it a sub project, he completely ignored me...
 
What a swell guy! But seriously, that's how support gets dropped for your project. You have to respond every once in awhile. Let us know that what we're doing is worth our time. I could care less if there are regular updates, just respond to our questions from time to time and I think we can all agree that that makes it worthwhile to contribute.
 
I concur totally. That admin just bites off more than he can chew. He didn't have the hardware/network to support things. He later got help elsewhere for that. Then he had the NCI app that did nothing and was supposedly going to get that working. Never did. Then mentioned another sub project that I don't believe ever happened. Then he brought up a BOINC stock market idea where you could buy points and that got shot down by most people. He hasn't fixed many of the issues over the last year and the project kinda just floats with terrible response time. Quite frankly I'm glad I got my goal completed there before all of the hassles came along.
 
Gilthanis, I passed you in Leiden :) and we're about to take a spot today I believe.
 
Indeed :) I'm thinking of staying on it for another week or so as we're only 2.5 and 5.5 days away from two more spotss, and those teams aren't particularly active so one part time machine should be enough to hold our position there.
 
Yeah...I have plenty of part time borgs attached...that shouldn't be a problem. Or at least until I reach my 1 million point goal. ;)
 
Just switched over to Leiden Classical to help us overtake those few positions!
 
Well NFS may not be a good idea on a MP rig depending on which WU's you get it can use allot of memory I have seen up to 2.1GB per core. One of my AMD 4P's is using 32GB of memory and 30GB of swap. Most of the time it is not to bad 20GB to 29GB and 1 or 2 GB of swap, but if I get the wrong mix of work it will bring the rig to it's knees. :(
 
Grandpa_01 said:
Well NFS may not be a good idea on a MP rig depending on which WU's you get it can use allot of memory I have seen up to 2.1GB per core. One of my AMD 4P's is using 32GB of memory and 30GB of swap. Most of the time it is not to bad 20GB to 29GB and 1 or 2 GB of swap, but if I get the wrong mix of work it will bring the rig to it's knees. :(
Click to expand...

The 16e work units are the big ones. If you are having troubles, just run the 14e and 15e work units and see how those turn out.
 
Yeah I know, the problem is with the 14e and the 15e they are not worth the power to run them compared to other projects, so I think I am going to run something else on the MP rigs untill I get some more memory.
 
Kai said:
Just switched over to Leiden Classical to help us overtake those few positions!
Click to expand...

:)

Looks like we have two more quick spots to take in a predicted 0.9 and 3.5 days, after that I'll leave a machine at least part time there and focus elsewhere. That leaves us in a position where if we want to, we can focus again in the future to take another 12 places without too much drama before the gap to the next team gets too large to make it worthwhile without some more firepower.

I've had some more wranglings with my eletricity company who are really beginning to aggravate me, to the point that if they do the same thing to me next quarter I'm willing to switch to literally anybody else just to not have to deal with them any longer. Price isn't the issue it's just an aspect of their service which is frankl;y ridiculous. Anyway, assuming I'm reading the meter properly, looks like DC costs have more than balanced out heating costs over the last quarter as we haven't run the heaters since the start of the quarter, and our electric bill is actually down on last quarter! Our storage heaters are incredibly inefficient - last quarter we used 1500 units on overnight heating! - so while not surprising, I'm pretty happy with that.
 
Sunday April 6 WCG will be doing some server maintenance. So, don't wig out because you can't get work, can't upload work, can't visit the site, etc... etc...

I would recommend that unless you have a backup project set up, to raise your cache so that you have enough work in the event that the server is down for a long period. It should just be routine maintenance though.
 
Wow, barely 17K ppd? Seems a bit low, doesn't it? Or is that just normal for the opencl on nvidia?

Disclaimer: only ever ran this on an ATI 4850...

Gilthanis said:
MilkyWay@Home v1.02 (opencl_nvidia)
Click to expand...
Gilthanis said:
Run Time 541.05
CPU Time 57.27
Points 106.88


Milkyway@Home Separation (Modified Fit) v1.28 (opencl_nvidia)
Run Time 1,933.51
CPU Time 646.50
Points Pending

Card info: http://www.newegg.com/Product/Produ...a&cm_sp=&AID=10440897&PID=3891137&SID=rewrite
Click to expand...
 
You will find that dual precision sucks donkey wiener on nVidia cards. Milkyway requires dual precision. So, AMD cards perform tons better on Milkyway...

You will also find that when it comes to OpenCL, AMD cards tend to have better support. It will also come down to the programmers skills at coding. A lot of times they will code OpenCL for one platform (which is then in essence more efficient for that device) and then try to make it work on all others. So if they do their coding with AMD GPU's, it is more likely to favor AMD GPU's. Throw that on top of poor OpenCL support and yes...nVidia again will under perform. CUDA apps would be much more desirable for nVidia cards. However, not everyone wants to learn multiple programming languages.

CUDA is supposedly easier, more mature, and more widely known due to Universities already having learned and taught it. However, not all of the sciences are coming from CUDA influenced sources and that limits you to nVidia only.
 
The bugs which lead to the huge amount of failed GPU jobs has been fixed. We're currently trying to figure out what's responsible for higher runtimes on Windows systems.
I hope we can solve this next week. Afterwards, I'll start some internal tests before reactivating the public Test Server.
So I'm afraid there is no schedule for the release yet, please stay patient.
Click to expand...

http://boinc.fzk.de/poem/forum_thread.php?id=1028&postid=9572#9572
 
Jathanis said:
Guess it's new enough BOINCstats BAM! doesn't have it yet huh?
Click to expand...

Yeah...fresh off the press. BOINCStats wont have it until either:
1. the admin contacts Willy @ BOINCStats
2. the admin publishes permission for stats sites to publish

Also, the admin asks that you not run multiple machines from the same location. Like anyone will really listen since most DC'ers are point whores. :D
 
I wasn't paying attention and didn't notice since I was WCG light lately, but MCM work units were temporarily suspended on the 7th and are now slowly being added back into the que. So, this results in people getting mostly work from other selected sub projects or no work from WCG at all depending on your settings.
 
Farewell from knreed
Dear Volunteers,

I have been part of the technical team supporting World Community Grid for the past 10 years. The project has been exciting, rewarding, challenging, exhausting but overall an amazing experience to be part of. A great part of this experience has been the interaction with you, the volunteers, throughout the years. However, as we all move through life, the time comes to move on to new opportunities and challenges. After a lot of hard thinking and deliberating, and after avoiding making a decision for a few years, I have taken an opportunity outside of IBM and my role on World Community Grid will move onto someone new. I want to personally thank each and every one of you for your interest and support for World Community Grid throughout the years and for always demanding more from us. Without your commitment, support, and ongoing expectation of more from World Community Grid, the program would not be the success that it is today. It has been a fantastic experience.

I started working on World Community Grid during June of 2004. I was thrilled at the opportunity to work on the project because I had been a participant at Seti@Home and grid.org. I loved the concept and opportunity to help advance science. Since I am in the consulting arm of IBM, I expected to be on the project for a few months to a year at most. Instead, I was lucky enough to get to work on World Community Grid for almost 10 years. It has been the high point of my career so far and I have thoroughly enjoyed the opportunity to work on this project and work with the researchers, other IBMers involved in the project and with the volunteer community. Together we have done some amazing work and contributed to the efforts of many different research projects.

World Community Grid is well positioned for the next decade. Although there are certainly areas for growth and improvement, the research pipeline is strong, the core support team is talented, and the commitment from IBM for the program remains strong. I expect great things in the coming years for the program.

I will remain a part of the community – but as a volunteer now instead of as part of the core technical team. I have always used my 'knreed' id in my official capacity but have done the majority of my personal contribution under the user name 'The Aspens'. I will be retiring the 'knreed' account and continuing forward as 'The Aspens'. I look forward to continuing to watch the growth of the community and the advancement of the science in the coming years.

Thank you for your support and contribution throughout the years.

- Kevin Reed
Click to expand...

https://secure.worldcommunitygrid.org/forums/wcg/viewthread_thread,36659
 
Is anyone crunching Collatz or Milkyway on AMD with the latest drivers? I'm having an issue with screen flickering on 13.12 drivers, might try updating but I'm hesitant to try it in case it goes horribly wrong - I'm not the biggest fan of AMD's drivers.
I did have a 6950 on this system which had the flickering problem really badly, now I have a 6970 in there which was fine crunching Collatz, but now having a slight issue with Milkyway. It seems to be only when not under full load, so basically in between work units, and if I scroll down in Firefox or click between windows the screen flickers really badly.
I've disabled ULPS to see if it would help, but it seems to be the same.
 
Are you telling BOINC to leave one full core/thread free?
 
You must log in or register to reply here.
Back
Top