BOINC

EXT64 said:
Have they said what version of GROMACS they are using? I know some of the newer versions have some nice speedups and AVX (not sure of the gains). Scaling, as we know from F@h, will depend on the size of the simulation.
Click to expand...

Based on task information, they run GROMACS 4.6.3, while Core A5 has been using 4.5.x.

Based on GROMACs release notes, below feature was introduced for the 4.6 branch :
New x86 SIMD non-bonded kernels for the usual cut-off scheme, called group scheme and the new verlet scheme, use x86 SIMD intrinsics (no more assembly code):

SSE2
SSE4.1
AVX-128-FMA (for AMD Bulldozer/Piledriver)
AVX-256 (for Intel Sandy/Ivy Bridge)
Click to expand...
 
Anyone interested in testing Gerasim? http://gerasim.boinc.ru/ I haven't seen anything bad yet, but haven't thrown a lot of resources at it. I believe it is Windows only, so some of our big hitters probably wont bother. ;) We are currently 18th in our League at Formula-BOINC for this project and I am considering the possibility of suggesting it for inclusion at DC-Vault. This would be considered a math project I believe.
 
I've been running some of the new GPUGRID GROMACS WUs. Seem to be running fine, though they are definitely in a stage of rapid core development (I've got v8.43 units, seems like the new batch is 8.44, and perhaps 8.45 later today/week?).
 
PrimeGrid just announced releasing the new version of their GFN apps. They say that the OpenCL apps may actually be faster than the CUDA apps on nVidia GPU's. However, the OpenCL apps will use a full GPU core whereas the CUDA tasks do not. More details in their forums.
 
Travis Desell contacted me via PM at Wildlife with an update. He plans on getting started on the DNA work sometime this coming week. He did not clarify if he meant releasing the work supposedly getting prepared or if he meant merging of the forums or if he meant more prep work. Hopefully, it will be work units. If so, those whom have goals may want to be ready and be watching the forums for an announcement. I plan on soaking up work units while they are available if they actually do get released. Travis tends to make a lot of promises that take a while to happen. Cross your fingers. For those whom don't know, he is the admin behind DNA, Subsetsum, and Wildlife.
 
Gerasim has already responded to not add them to any challenges (including DC-Vault) until September or possibly later.

 
I added Boinc@Fiit to the all inclusive list since BOINCStats now reports stats on them.
 
Calling all climate change scientists

https://secure.worldcommunitygrid.org/about_us/viewNewsArticle.do?articleId=375

Summary
In response to President Obama's call to action on the Climate Data Initiative, we invite scientists studying climate change issues to submit proposals for accessing massive supercomputing power to advance their research.....
Click to expand...

They also apparently had a BETA test today... https://secure.worldcommunitygrid.org/forums/wcg/viewthread_thread,37035 and it is for a new project!!

Uplinger posted this:
This application uses SSE2 instruction sets. This makes the application 20 times faster. We did a scan and it appears that we had 40 computers that were active in that week that did not have Pentium 4 or above with SSE2 instructions. Basically to allow the application to run on these older processors, it would do WAY more harm to the speed of the overall project by not running SSE2. This application will not run on processors without SSE2 instructions.

Thanks,
-Uplinger
Click to expand...
 
Last edited:
People are beginning to report problems downloading work. I am experiencing the issue as well. :(
 
My first result from the GPUGRID MP test. :cool:

http://www.gpugrid.net/result.php?resultid=12899192

I was using 8 out of 12 threads on the 980x. Heads up: The time elapsed and remaining reported in the BOINC Manager is way off. :confused:
Also, I paused the workunit after looking at the time remaining :eek:, but later decided to let it run just to see when the unit would actually complete.
 
Excellent runtimes. It looks like they're coming along nicely. This'll be a nice boost for non nVidia users. Oddly, Atlas also has estimation problems as well with actual runtime being many hours off of the estimate.
 
WCG will be doing a small BETA for some new CEP2 work units today. https://secure.worldcommunitygrid.org/forums/wcg/viewthread_thread,37045

Uplinger also gave an answer to my question about system requirements for the new project they are BETA testing.
The initial memory requirements are going to be rather low. I would say less than 50MB. The bandwidth usage for this project may be large for runtime of a work unit. The work units we tested in this beta were small.

Remember this is just initial data and can change when we get closer to release.

Thanks,
-Uplinger
Click to expand...
 
I would have created a new thread for this but it effects 3 projects and possibly a fourth future one. This was posted on DNA@homes front page by Travis.

project unification
Hi Everyone,

So I want to try and start unifying the projects to simplify server updates, checking the forums, etc. Before I get going I want a little feedback as to what I'm planning so everyone is okay with it.

Basically, I'm going to create a new project (probably called "Citizen Science Grid") or something along those lines. Unfortunately, because people have different accounts on each project I can't do an automatic merge of the users and databases. So this means you'd have to create a new account on the new project (when the new project starts those will be the only forums for the site).

When the new project is up and going, I'll allow everyone to transfer their credit from their other accounts via a webpage (basically if you're logged in, you can enter in the user name and id of your account on citizen science grid, hit okay and it will transfer the credit over). So a couple questions about this approach:

1. will it break the credit trackers too badly? I'll subtract credit on the old project to 0, and then add the same amount of credit to the new account on citizen science grid.
2. I kind of want to keep credit for the different projects separate; if I keep different tables and export the stats of each project as it's own part of XML, will that cause any problems? Or would everyone rather just have one combined credit for each of the projects here (Wildlife@Home, DNA@Home, and SubsetSum@Home -- we'll probably be launching another one in the fall as well, so I want to get this done before we go doing that).

So, once I get some feedback, and if people don't have too many issues with this I'll set up the new project and get things started. Otherwise back to the drawing board. :p

thanks, hope to hear from you soon about this!
--Travis 1 Aug 2014, 0:27:01 UTC · Comment
Click to expand...

I will also note that DNA sent a few work units out too. :D
 
Travis Desell posted this as an update

Hi Everyone,

I've come up with what will hopefully be a good solution to the project unification. I've created a new project called Citizen Science Grid, which you can take a look at here (note it's very rough and i'm still working on a lot of the webpages to unify everything):

http://volunteer.cs.und.edu/csg/alpha/index.php

Since Wildlife@Home has been the most recently active project, I'm re-using the Wildlife@Home database for this project. Note that this project won't export statistics. Instead, credit earned by this project will be exported into the various sub projects.

So this means if you run Wildlife@Home only, you don't need to do anything. If you run any of the other projects, first make an account on Wildlife@Home (or on the Citizen Science Grid webpage) if you don't have one already.

I'm currently working on a webpage where you'll be able to link the account on Citizen Science Grid/Wildlife@Home to your accounts on DNA@Home or SubsetSum@Home. You'll need to enter in your user name and email for the account you want to link it to. Then when you generate credit it will be generated for that old account.

In a week or so I'm going to shut down the forums on SubsetSum@Home and DNA@Home, because they're going to just use the forum on Citizen Science Grid. That way all projects will be using the same forum.

Does that sound okay?

--Travis
Click to expand...
 
I added a link in the first post for the BitcoinUtopia badge/awards/accomplishments thread I just created. Yes, they have badges now.
 
Version 7.4.15 is currently being tested and supposedly has some fixes for ASICs.
 
I personally am not a big fan of the way the BitcoinUtopia credit system is set up in Boinc. Supposedly the Boinc credit system is supposed to be set up to create equality between machines but if you run Bitcoin you can receive many times the credit for the same effort on the users side of things. It does not seem very equal to me but then again Boincs benchmark system does not work anyway when it comes to the differences between Intel and AMD on some prejects, so I guess it really does not matter anyway, the credit system as a whole is a farce anyway.
 
It depends on how you look at the fairness. I understand your point on the AMD vs Intel issue which I believe is an old debate at WCG as well. However, the current system is supposed to be based on FLOPs as you already know. The arguments being made against BU and the amount of credits given specifically to ASICs (let's be honest, nobody is arguing their CPU or GPU point awards) is really not much different than when GPU's first hit the scene. I agree with many others that GPU's, CPU's and ASICs should be in three different categories altogether. But it is too late for that boat to sail. There is also nothing to stand on when people argue that ASICs are single purpose. GPU's are pretty limited too. Only thing really truly universal is CPU. So, where does that leave things?
 
I would also like to add that projects don't take into account other resources when calculating their work units point values. Should work units that require 1GB of Ram versus those requiring 128MB be worth the same amount? What about work units that are heavy on internet bandwidth? How about the ones that require tons of IO's to the disk drive? None of the other factors come into play with points but it is hard to argue whether more or less science is being done. What about projects that require special sensors? The problem with scoring is the fact that unlike FAH or most other projects, BOINC is the universal can be used for all client. And that makes it easy for stats sites to tally them all. What it doesn't do is make them all equal due to the impossible task of coding for every future option. If it weren't for BOINC, would you be arguing whatever client WCG came up with against FAH? BOINC isn't the problem but the way we are all used to doing things.
 
Last edited:
Most likely mixing things will only end badly for Boinc, FAH has already proven that and this is no different than at FAH I would be willing to bet that in a relativity short amount of time there will be very few CPU's left and allot of miners which in reality do no science. I would say the majority of the current participants are there for the team comradery the competitiveness nature of the human race and the good you are or may be doing at the same time. Remove any one of those and you start removing participants.

I am pretty sure FAH (Vijay Pande) is regretting his decision to start the Bigadv program and the points disparity it created right about now. His active membership has dropped significantly and he is relying more heavily upon corporate donations and super servers now which come at a cost, because of it and the way it was managed.

People can say points do not matter all they want to but the evidence points strongly in the other direction, in any endeavor where a credit system is involved.

Ask yourself why are you or anybody else running it. ? I am sure it is not for the science value. ;)
 
I started running projects because I saw a huge wasted potential for idle systems. It later grew to the love of the science. Then morphed into competitiveness. And then has evolved into social structure with potential of real results.

But that is just me. :) However, I do agree that there is no happy fix. We just get used to it and keep on trucking. My biggest suggestion is just compare what you want to compare and ignore the rest.
 
Gilthanis said:
I started running projects because I saw a huge wasted potential for idle systems. It later grew to the love of the science. Then morphed into competitiveness. And then has evolved into social structure with potential of real results.

But that is just me. :) However, I do agree that there is no happy fix. We just get used to it and keep on trucking. My biggest suggestion is just compare what you want to compare and ignore the rest.
Click to expand...

The problem is it does not work like that we only have the credit system to compare to, the whole system not just parts of it, and Boinc will reap it's rewards or suffer from it. Alienating members has it's downfalls but that is up to Boinc, in reality the Boinc system is not set up all that well they have problems ranging from project management to credit system and most all points in between, they could learn from FAH's mistakes or not but that is up to them.
 
But whether to use the built in point system or create their own is up to the projects. That is part of the point of BOINC is to just allow a platform to be used by any wanting to. It shouldn't be up to David Anderson to dictate how someone else's project scores. Just as it isn't his place to dictate how the stats sites report them or compare them. One could argue that FAH could have learned from BOINC's mistakes... However, everything brought up is philosophical. What they need is someone who can come up with an actual solution rather than complaints.
 
Well the complaints have already started and they will grow and grow it was seen in the FAH system, this is just beginning. One solution would be to move all current and future Bitcoin points to it's own category and not include it in the Boinc Projects total. Look at all the top teams all of them have made around 50% of there points in the last 28 days running Bitcoin Utopia and there ae only around 400 donors running them and they make up what 75% 80% ? of boinc credit. Many including myself would not will not stick around for that type of disparity. If it was not for DC Vault and FB there would be no reason to at this time. Actually FAH would be the lesser of the two evils when it comes to credit disparity.

That is ridiculous.
http://stats.free-dc.org/stats.php?page=projmain
 
Last edited:
That is kind of unfair to do that. You are saying all of BOINC or FAH. How about saying... "I will leave WCG for FAH" because that is truly a better comparison. If the stats bug you, then it is the stats sites that need to change. However, if you picked one project to fairly compare to FAH and its pros/cons as you should, then I think you will find less issues with the BU issue. I know there are people ticked about the points but the solution of ignoring those projects are easier than telling a private party they can't report the numbers. I think this is more of a selling point for the team to focus on one project the team approves rather than over all BOINC points. Just food for thought. Either way, I will continue supporting most DC projects as I have done long before ASICs tipped the balance or even GPU's for that matter.
 
Not unfair at all I went from 43 RAC World Position in Boinc to 176 RAC in the last 28 days Boinc is a project that keeps track of credits for all it's subprojecs. So when you get lopsided in credits given to a project you will have donors migrating to that project within the Umbrella or leaving. I personally will not run any Bitcoin type program too much corruption within that system for my liking so I would be more inclined to move onto another project where I was more completive and more fairly rewarded.

But it is up to David Anderson as you said he is the one that allowed it and their terms. He could have just as easily said x amount of hash marks = y credits and kept it inline with everything else. But he chose not to. Even WCG keeps it's extra credits outside the Boinc system.
 
Again, BOINC is NOT a project. You can pretend it is all you want. But it is only a platform/client projects can use. I get the world position issue. But that is created by stats sites. Projects can also refuse to export stats altogether. David Anderson could also say FU to points, but that doesn't mean anyone will listen to him. If he scraps BOINC today, it is open source and someone else would just make their own.

You choosing a certain set of hardware is your choice. If you chose to not adapt, then you get left behind just like any other evolving system. BA was a good example of that at FAH and as you said, they are probably regretting that decision. However, David Anderson isn't in charge of all of the BOINC Projects. So, unlike VJ, DA does not have to worry about an ASIC screwing with his project. It only benefits him to support the new hardware. Don't get me wrong, I don't like the idea of one day of crunching totally outproducing my contribution for 7 years, but that started to happen when GPU's came along too. It took me a while to afford that hardware. I still don't have a top of the line card.

There are already suggestions being made at BOINCStats appealing for a way to filter project stats for comparison. Possibly by lumping them into categories. But that is just delaying things. You already mentioned Fomrula BOINC and DC-Vault being the positive that helps keep you here. Those are just stats sites. So, why does it matter what BOINCStats says or Free-DC puts in one of the many lists it has?
 
LOL
You contradicted your own statement, you are running BitcoinUtopia why ? Answer = Because of the points

Gilthanis [H]ard|DCer of the Month - July 2013/July 2014, 8.6 Years

I started running projects because I saw a huge wasted potential for idle systems. It later grew to the love of the science. Then morphed into competitiveness. And then has evolved into social structure with potential of real results.

But that is just me. However, I do agree that there is no happy fix. We just get used to it and keep on trucking. My biggest suggestion is just compare what you want to compare and ignore the rest.
__________________
Click to expand...

We are all completive as I said before I am not interested in not doing science If I wanted to donate money I would write a check. Computers are my hobby and my donation to society and a way to be completive and have fun. Boinc stats are part of that and a way to gauge your's / mine / everybody's contribution. If people want to use mining for science then do so but in reality it is mining and belongs in it's own category or the credit needs to be = I myself will not choose to compete against miners I will simply choose to either shut them down or find something else. Like you said it is My Choice to run Computers. I Like Them :D

How much do you want to bet the most everybody on the first 4 pages are running BitcoinUtopia for 0 science value. http://boincstats.com/en/stats/-1/user/list/ I am not really into mining. :eek:
 
You contradicted your own statement, you are running BitcoinUtopia why ? Answer = Because of the points
Click to expand...

That is actually insulting Grandpa. I have mentioned several times in IRC that I was running the project to understand how it is working with BOINC and be able to help troubleshoot in the future. Every project has the option to run funding by generating their own Bitcoins just like BU is doing only cutting out the middle man. That is actually discussed at GPUGrid by donors currently in regards to Donate@home (the original mining BOINC project and sister site to GPUGrid). I even mentioned earlier today how I wished I would have set up a separate account to run it under but it was already done. But my participation does not affect or change where you stand in regards to the over all stats. Since BOINC is in its infancy of supporting ASICs, there is a learning curve to it because it isn't exactly the same as how the rest of the miners do things.

Edit: If it was just for the points, I would be doing over double the points at BU from the ASICs I was sent.
 
The statement was made to prove a point people including you will run it because it will pay well computers can not compete with it in the credit field which is what is used to gauge contrabution in any credit system. It makes Boinc's credit system completely unfair and they do post users credit for all projects. This is unfair to all the other projects under the Boinc umbrella and there will be some that do the same as I will. I do not wish to compete against ASIC miners when there is such a huge disparity between the two. It is pretty simple but then again the loss of 400 cores 4 GPU's is apparently pretty minor compared to 1 miner at least according to the Boinc Stats.;)
 
I agree that there shouldn't be a union of mining and BOINC like there is. It dilutes the system of the science. Initially it seems that bitcoin utopia is an attempt to use the BOINC network to mine. Yes it is for a great cause, and the money is donated to science projects, but what direction is it taking if it supports asics? Is BOINC going to be another mining platform now? I don't like it. I appreciated the seperation of mining and BOINC. BOINC is for science. I have nothing against mining, I mined with ASICS for a small period of time before realizing a few doesn't make much money, but mining appears to be encroaching on the spirit of BOINC, and that's disturbing.
 
I think it encroaches the spirit of BOINC that most crunchers are accustomed too, but not BOINC as it was designed and intended. BOINC was intended to be used for any and all DC regardless of intent as long as the DC didn't break the open source license. That is why neither David Anderson nor Berkeley expresses any authority over any of the projects outside their own.
 
I created a test account for anyone to attach to if they want to support a project anonymously for any reason. I still have a few projects to add it to and I plan on setting a pc up here soon so that the CPID will all sync together. For those not knowing how to attach using a weak account key:
Here is what you need to do for BitcoinUtopia
•To attach a computer to your account without using the BOINC Manager. To do so, install BOINC, create a file named account_www.bitcoinutopia.net_bitcoinutopia.xml in the BOINC data directory, and set its contents to:
<account>
<master_url>http://www.bitcoinutopia.net/bitcoinutopia/</master_url>
<authenticator>3744_c914d6f6b2d8c112c4a847e9e8a7b86e</authenticator>
</account>

The accounts name is HardOCPtest for those wanting to watch the stats or look it up at projects.


If any adjustments to the accounts need made for adding certain work or what not, just let me know via PM. I am going to try and set up a few profiles so that people wanting to use it have somewhat of a choice.
 
7.4.21 is now available for download. Here is the revision info. http://boinc.berkeley.edu/dev/forum_user_posts.php?userid=8

Preliminary Change Log 7.4.19 -> 7.4.20


MGR: Fix bug which sometimes made project description text in Simple View too wide to read.
MGR: Fix bug in initial layout of Simple View project panel.
MGR: Ensure that the advanced computing preferences dialog is wide enough to fit all its tabs.
MGR: Set background color according to skin in Simple View dialogs; this is a slight cosmetic tweak for checkboxes on Windows 7, which have a narrow border using the background color at their bottom and right side; also don't use magenta background color for debugging backgrounds unless specifically requested by setting a new define TEST_BACKGROUND_WITH_MAGENTA_FILL to non-zero.
MGR: Cosmetic fix: eliminate brief display of all tasks when switching from Simple View to Advanced View when Show only active tasks is selected.
client: if a project's app_config.xml has no errors, remove old notices.
client: display XML in app_config notices correctly.
compile fix for FreeBSD; from Steffen.
MGR: Remove non-ASCII Ellipse. I'm not sure the code chunk is even used anymore.
LOCALE: Update BOINC Manager template.
LOCALE: Line feed fixes?
locale: Update compiled localization files.
MGR: Fix localization of menu item labels in BOINC Manager menu on Mac for wxWidgets 3.0; restore use of mac-specific UTF-8 ellipsis character so as not to invalidate old localizations of "Preferences..." menu item.
Mac Installer & MGR: delete BOINC Manager's wxSingleInstanceChecker lock file in case wxSingleInstanceChecker failed to delete it (such as due to a crash.)

Preliminary Change Log 7.4.20 -> 7.4.21


Quick Version Fix
Build break fix for Windows.
Click to expand...
 
There will be a new option for app_config files added to BOINC v. 7.4.9 and newer

<app_config>
<project_max_concurrent>1</project_max_concurrent>
</app_config>

This will basically set the project to only run 1 work unit at a time from that project. So, if you only wanted one work unit from SETI while dedicating the rest of your resources elsewhere, you could without having to tweak the project preferences. This is especially handy if you need more than the traditional 3 (4 if you count default) profile settings that most projects have. People with farms or borgs will probably be the ones to appreciate this the most.
 
Distributed Data Mining just became my 25th BOINC project to reach 1 million points!
 
You must log in or register to reply here.
Back
Top