Separate names with a comma.
Discussion in 'Distributed Computing' started by Gilthanis, Sep 30, 2013.
The magic of BOINC! That and you can tailor projects that take less time to your slower CPU cyclers!
Just a warning, if you attach to Climate@Home (not climate prediction by the way) make sure your system support VTx. I believe this is why my Q8200 was BSOD'ing. However, I don't know if it was VTx related or just the project itself.
Good info. Thank you.
Finally have 20 projects attached! Time to let the WUpop badges flood in.
I broke the BOINC 500 million point barrier today! Keep crunching!!
Congrats DooKey! Its been a long time coming! Well done.
I see youhave been dominating Collatz.
Congrats DooKey! metallicafan, good to see you have returned to the forums.
Holy crap! Congrats DooKey! The rest of us can only hope to reach that!
Damn! I'm going to me happy to hit 50 million by Thanksgiving...
Way to set the bar for the rest of us! I guess it's time for more video cards - come on Black Friday/Cyber Monday!!!
^This - great to see you back around here!
Okay, I guess I haven't looked in a while, but kudos on the recruiting for DIMES! I mean dang, BigTed and Musky are putting up 6 figure daily totals, and Grandpa isn't too far behind!
Guess I'd better get on some VMs, or I'm gonna be left in the dust...
Why do I have negative points for my credit/day, /week and /month, and an average credit of 40?
This concerns me...
Not sure if I've set up something wrong but that doesn't seem right at all. It also seems to think I've dropped 24 places in the team in the last month, I sure don't remember being up in the top 5 oh and Asteroids@Home doesn't show up in my stats on my user page, but it does on my hosts' pages (well, most of them, some say the ID wasn't found).
Possibly a cross project ID issue. They recommend that you have one system connect to all the projects you are a part of so that it syncs better and unifies that cross project identifier. If one project doesn't match up, it can sometimes throw your stats off temporarily. That is just a guess though.
Well...I just got a borged Hex core dropped off to work on. Evidently the power supply has already crapped out. It was a 600W Ultra and the system wasn't pushing all that much power. Looks like I'm doing some warranty work over the next day or so. In the mean time, maybe I will slap my 400W backup into it and crank out some DIMES VM's...hmmmm
This came to me last night, but I guess part of it might be that I used the same name and email address previously for another team on Rosetta, but later changed my team on my profile when I moved here...sound plausible? So that should iron itself out I guess, maybe when I overtake my "other self"?
And, are you saying that I should set up one system that's attached to every project I have attached to in the past, to help it sync up? If so, that's not a problem, I can do that when I have a system again.
That is what I am saying. Otherwise the stats sites might conflict on whether you are the same person or not. The cross project ID is what identifies you at multiple locations. Don't ask me why it was done that way, I'm not that smart.
So just to be clear, this might not iron itself out until [H]CaptainUnlikely overtakes [OcUK]CaptainUnlikely, my former self, in Rosetta? That could take a while, I ran Rosetta for quite some time there! But, no matter, I'll get that sorted at some point.
I'll attach my machine to every project I run when I can, and see what happens, it's not a huge deal if it's just a stats glitch it just worried me slightly. Nobody wants to be going backwards
I'm not saying you have to overtake yourself at all. It has nothing to do with points. It has to do with identifying yourself to the multiple projects. Maybe I'm not seeing what is happening and thus not getting the problem. However, if you are showing negative points in your stats, it is usually because the stats site thinks that the account at the project is someone else. This usually happens when the cross project ID changes at one site but not at all of them. You crunching under another teams banner shouldn't affect this at all. You can bounce between teams all day long. The only thing that will do is show smaller team contribution. However, your total stats should still look normal.
Edit: Now if you have more than one account at Rosetta, you could also try asking an admin to merge your accounts. Some can do this and some wont. YMMV and I'm not sure what kind of trouble would result from doing so.
Edit2: Go to Rosettas page and login into your account. Scroll down to Cross Projects Statistics. Under all of the stats sites should be Cross-project ID: (and some numbers/letters). Then go to another project, lets say WUProp. Login into your account. Compare your ID from Rosetta to the one listed here. If they are the same, then verify that at the other projects. If one of them is different, then that is the problem. If all of your projects are the same, then I'm not sure.
A couple random notes:
You really need to watch which team you join on some of these projects. SIMAP is one - we have a HardOCP team and a [H]ard|OCP, and HardOCP is the one tied to us in DC-Vault. It would behoove you to check HERE first to make sure you get the right team.
Correlizer - is it just dead, on on a break? DC-Vault says we are doing poorly there, but there appears to be no work to do.
Does anyone have CLI Linux instructions for setting the BOINC client up? Install is easy enough, and I did find a list of commands and parameters somewhere. If no one does, I'lll try to write one up while I am setting this up in the next few days.
FYI, our DC-Vault ranking bothers me. The criteria used is crummy - #1 for a random now-dead project counts the same as #1 for F@H, which has to be exponentially bigger than any one BOINC project, including WCG (the biggest one?)? Given this, I'm going to try to manipulate the stats on these other projects with some of my horsepower, at least short term. I'll have a 2p socket 2011 machine coming up in the next few days, and I think I am going to put a 4p socket 2011 system towards this as well - still undecided on that one. Either way, hopefully I can make a noticable difference, especially in these smaller projects.
musky, I agree the Vault has some major issues. Unfortunately, I have been unable to register an account with their forums so that I could post and address my concerns. I have also tried getting ahold of the admin and have yet to receive a reply from them and I have sent them messages multiple times over several months. Correlizer was pretty active until August. I'm not sure what is going on there at the moment. I personally feel that Correlizer, ABC, and RNA should be pulled, while WCG should be reclassified under the Miscellaneous category since they do research in fields outside of Medical/Biological and have done so for many years now.
CaptainUnlikely, looking here I can see not all of your projects are listed because WUProp is not on there nor is Enigma and I suspect others ... http://boincstats.com/en/stats/-1/user/detail/2437003/projectList
Edit: musky, I don't have anything on setting up CL versions of Linux. I have never attempted it. I think a guide would be valuable to others if you could set that up. Also, a guide on how you got DIMES running on Linux would be awesome too.
Edit2: musky, I also address the team name differences in the BOINC setup guide here: http://hardforum.com/showthread.php?t=1768558 I did this since not everyone will have desire to participate in the DC-Vault.
not 100% on this but i think musky came to the conclusion that setting up DIMES on linux (CentOS) required too much ram vs using XP. something like 1gb for Linux vs 192mb in XP. but i will let musky speak to that
Yeah, but it would still be good for if anyone was already running linux as their primary OS. I agree as a VM it would be poor choice.
I would recommend that you also attach to WUProp. It is an NCI app that will run alongside other projects. They also have badges and you can search their results for some helpful information.
As far as the point awards and popularity concern... A less known or liked project is easier to get points because of lack of competition and that the project should not be penalized due to popularity. It is for the science after all. I for one don't feel that mathematics and bio sciences are equal. Nor do I feel cancer research will serve more people than something like clean water in 3rd world nations. So when it comes down to the points, the popular ones are indeed harder to achieve #1, but #1 is still worth the same points.
Edit: SETI has the most registered users at 1.4 million+ according to http://boincstats.com/en/stats/projectStatsInfo. WCG is second with only 418k+. Einstein has the most devices at 5.9 million + and SETI comes in second at 3.4 million+. However, as mentioned above, I agree that the Vault certainly has its problems/flaws.
Though I haven't done a lot of research and comparison, according to FAH's page here: http://folding.stanford.edu/home/ They have less then 300k devices.
Gilthanis - thanks for the info, I noticed not all of my projects are listed so I will go and check my CPIDs, my ID for Rosetta looks OK at least. I only have one account, I just changed teams, I got confused on what happened as it was a while ago, sorry.
I was on the wrong team for SIMAP though, d'oh what a nooby mistake. I thought I'd checked before joining to see the active members, I usually do, must not have done for that project. Corrected that, anyway, for next time I run it.
musky - I recently set up BOINC on my CLI Linux box, I could have a go at writing a guide but I'm definitely no Linux expert so it might not be exactly right although I did find it quite simple even for me. I'm sure if you did have time to write one, it would be useful and appreciated. I would also be very interested to see how you got DIMES working on CLI Linux.
Easy - I didn't....
I got it to work on CentOS whatever the current version is, which appears to still use the Gnome2 desktop. The problem is, that does us almost no good since no one uses a GUI version of CentOS for a DC box.
I am noticing that here is your other projects. http://boincstats.com/en/stats/-1/user/detail/2634642/projectList I'm guessing BOINCStats thinks you are two different people due to the cross project ID issue. Even if you attach and assign the other projects to no new work, it should in time get the cross project ID sync'd up by attaching one PC to all of your projects. After it gets sync'd up, you shouldn't have to do that again unless you join a project for the first time with no other projects attached to the device.
One of the things I do is to setup up BAM! to automatically attach me to WCG and WUProp on every device I attach to. Then I add other projects from BOINCStats and do an update. This ensures that every device starts out with the cross project info from those two sites. Just a suggestion.
I went ahead and added every project I've crunched onto my laptop last night, and set everything but Asteroids and WUProp to no new work. I set BAM to attach new hosts to WUProp by default anyway, which I now realise was a good idea for two separate reasons as that should now keep this from happening again when I attach new hosts. I understand the whole CPID situation a little better now, my brain was totally fried last night so I apologise for my density heh. I will just wait for it all to sync up now
No need to apologize. There are many times I go to help someone and am way off the mark. Hopefully, this will rectify things.
Looks like my stats issues are now resolved, thanks for the tip
Here is another non-CPU intensive project. It is new and still in its BETA stage. They currently only have a Linux version as well. I have sent Razor_FX_II a message to create a team name for us.
I have started up my two Ubuntu VM's to test it. Supposedly it uses heavy network bandwidth, but that is all the details given on their forum. I don't have any numbers to support what constitutes "heavy".
Looks like an interesting project. Does Ubuntu have "GLIBC_2.14" which I see listed as a requirement, built in already, or does it need to be installed separately?
I will go ahead and attach my Atom to this when we have a team etc, mainly to gauge whether it has an impact on other projects on a low powered system.
Yafu now has an Android app. http://yafu.dyndns.org/yafu/forum_thread.php?id=143
CaptainUnlikely, I think my clients were updated to version 12 (Ubuntu) and I didn't have to install anything.
More info on plagiarism@home requirements http://plagiarism.boincpolska.org/plagiarism/forum_thread.php?id=9&postid=45#45
Is it typical for various projects to run out of work to do? I fired up malariacontrol.net on an Intel 4p machine this morning at around 9 AM. By 4:00, most of my CPU threads were not doing anything. Is there some setting I need to change to make the BOINC Manager send work/get work more often, or am I simply going to need to run more than one project on it to keep it busy? I added SIMAP a few minutes ago, and it is fully loaded once again.
Also, a fairly alarming number (close to 100) of malariacontrol tasks showed a status of "Computation error". This machine is a stock speed 4p E5-4650 server. The chips are "extra spicy", but they are a later C0 stepping and have been flawless on Folding@home for several months now. Is this something I should worry about? I have a similar number of "Ready to report" jobs from this project waiting for upload, if that matters.
You can check their server status link on their BOINC page to see if there is work available. You can also see if the feeder is running. This would be a good indicator of work. Usually if I see a lot of errored units, I will go to their forums and see if anyone else is having issues. Sometimes there will just be bad batches. I haven't ran much Malariacontrol lately, but I could take a look a little later.
Also, if you have many failed work units, they may reduce the number of work units you can acquire in a day until your reliability rating goes back up. Check your logs to see what the message from the server is.
I don't see any errors listed on any of the 3 systems you have. Perhaps they have already purged those results. http://www.malariacontrol.net/hosts_user.php?userid=105196
And to answer your first question, some projects do run out regularly while other projects seem to have an endless supply. SIMAP runs out regularly from what I recall. Correlizer has been dry for a few months. ABC has scattered showers. It is best to watch how many work units are still out in the wild. When they get returned, more are typically generated. I had to cache as many as possible when they became available in order to hit my goal there. POEM runs out of GPU work units pretty regular because they can't supply the demand. However, there seems to be CPU work quite regular. But that is just covering most of the problem childs of DC-Vault. There are various other projects outside the Vault that also are plagued with various issues. I find it best to always keep a reliable project like WCG attached (even if you give it a priority of 0) to guarantee cores/threads don't go dry.
knreed @ WCG has given a small update on the new cancer project. https://secure.worldcommunitygrid.org/forums/wcg/viewthread_thread,35527_offset,40
Galaxy S3 sitting at my door. When I get home, 2 more cores for crunching.
Can't you just register on the Team Ninja forums? I'm pretty sure thats all I had to do to get any account although its been several years ago now since I did this.
I'll try and make a post about Correlizer on the forum tomorrow when I have a bit more time. ABC doesnt have a lot of work but I did get some the past couple days. Judging by the stats RNA puts out some work just not a lot.
Also ... Great works on DIMES guys! You all are making great progress there. Jathanis - nice work on Android@home. That project is scoring very well and we have a chance to gain some spots there if anyone else wants to come help!
Hey, just trying to keep pace with you! F@H isn't the only place for friendly mowing competitions! Seriously, I had first just thought about getting another project to 1M, but I really like the project, scores well for CPU work, and my daughter loves the BOINC screensaver it does on her computer... Might have to stick with it until the annual WCG December competition.
The answer is No. I have asked another member of the forums to get ahold of Rusty. Something is wrong with the site. 4 different email accounts from 4 different providers all come up saying that email account is banned. Even messages to the webadmin go unheeded. The other member that is contacting Rusty for me even attempted just in case I was being incompetent. He couldn't register again either. So, we wait.
RNA certainly has work. My issue is that I don't believe it is acceptable for the Vault in its current and past state. Short work units are in very "short" supply. Long work units (until the new VM versions are final) don't checkpoint and run for weeks. The new VM versions are supposed to auto extend the deadlines which was another huge issue. And the fact that they are a BETA project alone tells me they shouldn't be in the Vault either.
Correlizer just doesn't have work, but runs good when it does. ABC doesn't have steady consistent work any more which is one of the things that qualifies a project for consideration. SIMAP also has a supply issue but not like the 2 others mentioned. POEM has plenty of CPU but is terrible about keeping up with GPU. However, that isn't a disqualifier.
WCG is in the wrong category. It should NOT be in the BIO/Medical section. They aren't just bio/med, but rather they are humanitarian. Since they have had multiple non-bio/medical projects, it should be like YoYo and put under Misc. Examples of non-bio/medical... African Climate, Nutritious Rice for the World, CEP, CEP2, C4CW, and CFSW
Oh and add Spinehinge on the dead project list. Certainly not worthy of the Vault these days and should be pulled.
I also feel that if they are adding POGS, then they should also add theSkyNets Source Finder. It runs solid and is a lot easier than what the forums are giving them credit for. The only trouble is how they tried to merge the two project when it comes to teams. If you were joined at both teams made, they automatically migrated you over to the pogs version. That is problematic since they are combining points because since Phoenicis hasn't joined our POGS team, he is a lonewolf in the old one and thus his points don't cross over to us. That is just a minor irritation. I have asked Bok to find a way to add the Source Finder stats on Free-DC because some people may choose not to run POGS.
Oh...and what is Android@home? I must have missed that one. Got a link? Or did you mean Asteroids@home? The two of you are doing awesome over there.
musky, I finally saw one of your error work units. This is the error and please correct me if I'm wrong: http://www.malariacontrol.net/result.php?resultid=183486261
process got signal 11
If my memory serves me correct, that was a known issue at some projects with the version of BOINC in the Ubuntu repository for version 7.0.24 and I believe also 7.0.27. I would suggest upgrading your BOINC install to a later version. See if that helps things.
My gf's laptop errored out 10 Asteroids units with the following:
-226 (0xffffffffffffff1e) ERR_TOO_MANY_EXITS
Seems to have been doing alright up until then, and since, so maybe a fluke?
Edit: also, I'm down a Sony Xperia P (dual core 1GHz Android) as I sold it, but on the plus side I'm borging two part time C2D laptops this weekend.