DC Vault 2

SETI is evidently down for 3 days to migrate their servers. Oh well...Milkyway and Albert is still chugging along.
 
Evidently Climate Prediction is having troubles with the newer BOINC Alpha clients. 7.0.28 is supposedly working fine. I can contest that the Alpha clients up to 7.0.59 are suffering from the MD5 download issue. So, for those looking to run Climate Prediction or upgrading clients, beware. I do occasionally get a successful download, but several fail. I don't plan on downgrading, so I will contribute what makes it through and will rely on backup projects to keep my systems busy in the mean time.

http://climateapps2.oerc.ox.ac.uk/cpdnboinc/forum_thread.php?id=7560#45736
 
Thanks for the heads-up, I'll stick on 7.0.28 to avoid any issues. Is there a specific advantage to moving to the alpha client?
 
The app_info files get replaced with the much simpler app_config feature. The new app_config gives you a lot more control over what and how you crunch. Like if you were supporting WCG, you could tell the client to only run one CEP2 work unit at a time in a multi core system but still have CEP2 work units cached. Where as without the app_config, you would only have the option to download one CEP2 at a time if you didn't want more then one running. It will also do everything the app_info files did. If you don't use these, then I doubt you see much change. There are different things going on under the hood, but I doubt many would ever notice. I am slowly updating my systems to 7.0.60 in hopes that it fixes the Climate Prediction issue. Borged systems will just have to wait.

Also, WCG plans on eliminating the app_info option in the future in support of the app_config feature.
 
That does sound interesting, potentially very useful. I'm not exactly an advanced BOINC user as yet though so I'll hold off on the upgrade until it's a mainstream release.
 
Well CEP2 is hard on a system's hard drive with all of the I/O requests it does on the disk. So, running too many on one system using a traditional hard drive may actually cause:
1. The computer to be super slow and unusable
2. lose efficiency due to the work units always waiting to write data
3. Crash your system (some hard drives can't even sustain one work unit)
Edit: 4. Cause other WU's to crash due to taking too long to write to disk.

So, on an i7 capable of running 8 threads, this can be a nightmare. However, I have seen some systems that did all 8 just fine. It all depends on your setup. You will also find some people use Ramdisks for this reason. You can limit the number of CEP2 work units on WCG's page, but that also means you only download that many at any given time per computer as well. The app_config lets you download enough CEP2 to fill your cache for the number on concurrent work units you have set the system to run. It is extremely helpful in that case. You will find that other projects outside of WCG also have some pretty hefty requirements. Some projects want over 4GB of RAM per work unit. On a mutli-core system this would put the system into shock with all the caching in virtual memory. I would bet it would cause the work units to fail unless you had 4GB's of ram per core in your system. So, by using app_config, the user could tell their client to only download one at a time and the project would not have to put limits on their side.
 
Last edited:
Asteroids has been added to the vault. We are currently in 8th place with 9,822 Vault points.
 
Also a shout out to the WCG team. We have exceeded 2 million WCG points for the past 12 days in a row. A feat that I'm fairly sure has never been seen outside of the Christmas Race. For a reference point at this time last year our daily totals were around the 600K - 800K range. Props to all of you out there contributing to the great research being done. We have several new members and its exciting to see the team doing so well. Keep crunching for science my friends.
 
Yeah... GPU crunching was attributed to ending HCC 2.5 years early. I would like to see if we can keep our numbers in the 2 Million mark after it ends. I'm betting GPU is where most of the points are coming from.
 
I moved all my boxes over to HCC a couple days ago. Should have 5 years runtime before the project ends. No GPUs here. Have 30+ cores spread across several machines.

Kudos to the WCG crunchers out there! Keep up the great work! Nice to see new members on the WCG team.
 
Last edited:
I sold off my 1155 Celeron, board and RAM, and I'm on my Core 2 laptop temporarily until the SR-2 is online. Crunching Climate Prediction on one core, gets a bit loud if I crunch on both, still running WUProp and FreeHAL though. Set MrsUnlikely's laptop up on Asteroids on one core and WUProp (only has a small SSD, no room for Climate Prediction), and my parent's quad Athlon is now on Climate Prediction and WUProp. I also finally got the tablet I've been waiting for that got delayed in the post, it's 24/7 on Enigma now. 1.6GHz dual core, be interesting to see how it does.
 
Enigma is the highest point rewarding project available to nativeBOINC right now, so it will probably do well. Don't forget to post your badges! I have my phones attached to each of the available projects. I think Subset Sum is the next highest scoring project. I would recommend having it set as a secondary project just in case Enigma goes down or runs out of work. Unless, you want to stay focused on the Vault. In that case, just choose one that you like from the Vault projects that support nativeBOINC.
 
Thanks for the tip, I'll add Subset tonight so they don't sit idle if Enigma goes down. Haven't earned any badges on Enigma yet, but I'll post them when I do!
 
These are the nativeBOINC projects that are in the Vault.
Asteroids
SETI
Enigma
Milkyway
PrimeGrid
YoYo

Out of those projects, I think Milkyway comes next after Enigma in points. However, I don't have any exact science or numbers on it. I would reference the stats on www.nativeboinc.org for more details.
 
WUProp is also supported now, right? I saw it yesterday on the list, guess I may as well install that too.
 
Yes it is supported, but I didn't mention it because it wasn't in the Vault and I assumed you already attached...my bad.
 
Ah, I didn't realise it isn't in the Vault, my mistake. I haven't attached just because I didn't see it when I was setting my phone up, only saw it when setting up the tablet so I must've missed it first time around.
 
Just an FYI
Since updating to 7.0.60 clients, I have not had the MD5 download error at Climate Prediction.
 
For those running Climate Prediction, there has been another hiccup. Some work units are getting a Fortran error. These are bad work units. I would recommend aborting the ones that do this so that you can get the new work. Trying to finish them would be a bad idea because they pause and don't actually do anything.

http://climateapps2.oerc.ox.ac.uk/cpdnboinc/forum_thread.php?id=7564#45832

I'm just hoping none of my borged machines get this error...what a nightmare that will be for home tech support. :p
 
So, what would people like to crunch for the month of May? I was thinking Correlizer for CPU and Milkyway for GPU. Any thoughts?

On a side note, I got another part time borged dual core added to my list and possibly a laptop later this week. I even swapped out somebody's single core rig to a dual adding an additional core that will be running 24/7.
 
Due to my request, OProject and PrimeGrid have added the badges for each contributor under our team list so that we can easily see the achievements of our team mates. Check them out!

http://www.primegrid.com/team_members.php?teamid=1710&offset=0&sort_by=expavg_credit
http://oproject.info/team_members.php?teamid=1670&offset=0&sort_by=expavg_credit

YoYo, WUProp, and Numberfields already do it. WUProp and NumberFields also did it per my request previously. It is nice to see active admins working on suggestions.

Well...since CPDN continues to have bad work units and they refuse to send a server abort for whatever their technical reasons may be, I had to detach many borged machines. Started to get phone calls worried about the sudden error they couldn't get rid of.
 
Last edited:
Gilthanis said:
Anybody interested in getting Intel GPU running? I would love to know more details, but I don't have an Intel GPU capable at this moment.

http://setiathome.berkeley.edu/forum_thread.php?id=70717#1350603
Click to expand...

Sounds interesting but I don't have any IB processors and am unlikely to in the foreseeable future. Neat that someone's finally going to put them to use though - I always thought Intel graphics (the recent ones anyway) were like phones, maybe not very powerful for DC but there sure are a lot of them out there, so potentially a good source of compute power.
 
It's still neat that Intel is starting to support it.
It may be a small step, but at least it's in the right direction, and if it helps just one person, that's what counts.
 
Well...I'm fed up with Climate Predictions constant headaches. I am still getting some of the MD4 download issues, still getting calls about the Fortran issue, and am now seeing C++ errors popping up because of their work units. It seems ever since they got new staff, they have been plagued with nothing but trouble and their attitudes towards fixing things seems to be geared towards Linux users. So, I will set my clients to "no new work" and just move them somewhere else. Maybe if they get their act together I will return.
 
Nominations are almost done for Hard DC'er, so get them in now if you are thinking of anyone: http://hardforum.com/showthread.php?t=1755626

We also haven't heard any other suggestions on next months focus. I nominated Correlizer for CPU and Dookey seconded it. Any other thoughts? What about GPU? I suggested Milkyway since it has both AMD and nVidia work. (POEM keeps having a supply issue) The only problem with it though is that it requires dual precision cards. So, anyone using older tech will not work out there. I have a few of those cards, but they will just do work where they can.

I also passed 80 million points BOINCwide
 
Last edited:
I'm happy enough to switch over to Correlizer for May, I don't have much to throw at it but I'll move over what I have. I have no GPUs at the minute so I have no opinion on that side.
 
Guys, I'm going to take a hiatus from BOINC for a bit and turn my attention back to Folding. I have a couple of goals to meet there, but I'll return to BOINC sometime in the future.
 
Give 'em hell Folding, DooKey - you'll be missed over here, though! Dmaine & I will have to hold down Primegrid until you get back... I've been eying some 660TI/670/680s for Folding again once my funds allow, but the little bills seem to be keeping up with my efforts to save...

As for May, I 'm just working on WCG badges for the rest of the month, but I'm game for Correlizer on CPU for May. And as I don't think any of my cards will run POEM, I can try the ATI over on Milkyway again - it got me 10M there at some point in the past, so unless they changed the client, it should still work... In fact, I might be able to pick up a few places on our team, too! :D

And FWIW, I should pass 30M total BOINC this weekend... Need moar GPU power...
 
Jathanis, my 210's crunch at POEM ok, so if you have nVidia cards I don't see why they shouldn't work. My ATI 4350 certainly doesn't though. (I didn't expect it to since it only had BETA support) Which ATI card are you running?
 
DooKey said:
Guys, I'm going to take a hiatus from BOINC for a bit and turn my attention back to Folding. I have a couple of goals to meet there, but I'll return to BOINC sometime in the future.
Click to expand...

Best of luck with your folding goals Dookey. Your BOINC contributions over the last couple of years have been simply outstanding and it's been a pleasure to crunch alongside you.
 
You must log in or register to reply here.
Back
Top