Projects where the team needs help

doneske

Well-Known Member
USA team member
I have decided to freelance a little bit after crunching almost exclusively at WCG for the past 16 years. Are there any specific projects that the team needs some additional computation power on to reach a goal or to build ranking? This will be exclusively CPU power, no GPUs.
 

Nick Name

Administrator
USA team member
If it's not Rosetta, Asteroids, Universe or GPUGrid, it's probably a good candidate. :D I think one of the more interesting projects to come along is MLC@Home, a project researching machine learning and neural networks. We've only got three members crunching there so far, so can definitely use a boost. You should feel right at home there as it runs well on Linux.
 

doneske

Well-Known Member
USA team member
Coincidentally, I started attaching my farm to that project (MLC@Home) yesterday around noon. Have moved a few more machines over this morning but started receiving "no tasks available". Hopefully, it is only be temporary. The big machine will be the test as it will probably need about 500 WUs minimum. My short term strategy right now is to run MLC until team is in the top 10 or so then move to TN-GRID until team is at least top 20. After that will look at the rankings and pick another project from there.
 

Nick Name

Administrator
USA team member
I haven't seen many reports of not getting work at MLC yet but I don't know if anyone's put the kind of power on it that you have. :D TN-Grid is another interesting project, we've never had much team interest there. You may have trouble keeping machines busy there, work generation is slower than other projects there due to the nature of the work. They haven't actively promoted the project because of that.

Just a note, you have to consent to data export for the stats there to count. Log in -> Project Preferences -> check the consent box for export.

There is the upcoming PrimeGrid challenge if you're interested.

Summer Olympics Challenge has been replaced with the Katherine Johnson Memorial Challenge.
TRP-LLR challenge from 24 July 20:00 UTC to 31 July 20:00 UTC.

QuChemPedIA@home is another interesting project, working on molecular chemistry. We've only got one active cruncher there now.
 

doneske

Well-Known Member
USA team member
One can get work on most scheduler requests but it may be only 2 or 3 WUs at a time which is interesting as they always seem to have about 80000 ready to send. Team USA has now moved into position 16 from 27 yesterday. I still haven't moved the entire farm to MLC yet as most machines still have work draining from WCG. The big 128 thread machine is still 100% WCG and will be for at least 2 more days. Right now, I'm just interested in moving the team into the top 20 on most projects. On small projects, I may split cores between projects and crunch more than one concurrently. I'll take a look at QuChemPedIA@home as a possible target after TN-GRID.
 

Nick Name

Administrator
USA team member
That sounds like something's not set right on the server. They've been working out the bugs, that sounds like another one they should look at. The server should send as much work as the client requests if it's available, unless they've set a hard limit on how many you can have like SETI did.
 

doneske

Well-Known Member
USA team member
MLC@Home was able to maintain work for all machines yesterday so there is enough to go around although a little uneven at times. Team USA has now moved to position 12 from 16 as of this morning. Another solid day of crunching should move us to the top 10.
 

doneske

Well-Known Member
USA team member
I've been quoting the RAC ranking from the individual project site. Maybe a better metric would be the ranking from Free-DC that goes by cumulative points. RAC is just too fleeting to me. Once you stop, the RAC drops rather quickly. Do we have a standard metric that we use?
 

Nick Name

Administrator
USA team member
I've never delved too deeply into that. I know there are different ways to calculate it, some projects it's more accurate than others. MilkyWay in particular shows a high RAC for weeks after you stop crunching. Like you said, Free-DC has its own method (the FDC RAC column) and that number can differ quite a bit from what the project is reporting. I don't know how Bok is calculating that. I've always used the project figure as it also closely matches what BOINCTasks reports. For sure, the individual project RAC rank is based on that project's stats, so it makes sense to me to use that number.
 

doneske

Well-Known Member
USA team member
Moved over to TN-Grid and have finished about 2000 WUs so far. Work seems to be steady and all machines are full. Last time I looked team was ranked at #17 according to the project stats and I'm trying to move us up into the top ten.
 

doneske

Well-Known Member
USA team member
The team has now moved into 9th place on TN-Grid. Starting to drain down in preparation of moving to QuChem. That should be an interesting project from both an availability standpoint and type of work. From what I have read it looks like it might be a VirtualBox project which isn't one of my favorites due to the various issues surrounding the use of VirtualBox.
 

Nick Name

Administrator
USA team member
The team has now moved into 9th place on TN-Grid. Starting to drain down in preparation of moving to QuChem. That should be an interesting project from both an availability standpoint and type of work. From what I have read it looks like it might be a VirtualBox project which isn't one of my favorites due to the various issues surrounding the use of VirtualBox.
I didn't think it would take long to move the TN-Grid needle. :) Good news for you, QuChem runs natively on Linux. VBox is only required for Windoze. :D

[Edit] I forgot to mention QuChem requires an invitation code to join. It's on the front page: 3VwMu3-eTCg32 [/Edit]
 
Last edited:

BeauZaux

Well-Known Member
USA team member
On small projects, I may split cores between projects and crunch more than one concurrently.
I tried the split cores a few time, had problem when a random big WU dropped in. I know better now on controlling that, but wonder what projects you would recommend for split coring.
Thanks
 

doneske

Well-Known Member
USA team member
I normally don't run too many (more than maybe 2) BOINC projects at the same time. I find that the work is too varied and something will eventually step on something else. Plus, BOINC has a difficult time trying to manage that diverse of a workload. My plan was to run QUChem, and if work became scarce, I would add TN-Grid back in to fill in the "holes". However, QUChem has long work units that are multi-threaded so that isn't an option. I guess I could inhibit the "long" workunits in the preferences and/or use an app_config.xml to manage the concurrency. It's just a matter of how much work and baby-sitting I want to do. QUChem is still a small project so it won't take much again to "move the needle" so I may just move on to another project if work becomes too difficult to get. Right now, the work availability is fine and the work is long running so it shouldn't be much of a problem.
 

doneske

Well-Known Member
USA team member
May have to reevaluate my participation in QuChem....

Secondly, I am very busy and don't have time to answer the same questions/problems many times. This forum takes me a lot of time for trivial problems (points/pending) or problems that have already been answered. We don't have any technical staff or trainees to solve some known problems. Boinc project is a tool to help researchers and not to divert them from research to satisfy impatient and lazy participants.
 

Nick Name

Administrator
USA team member
May have to reevaluate my participation in QuChem....

Secondly, I am very busy and don't have time to answer the same questions/problems many times. This forum takes me a lot of time for trivial problems (points/pending) or problems that have already been answered. We don't have any technical staff or trainees to solve some known problems. Boinc project is a tool to help researchers and not to divert them from research to satisfy impatient and lazy participants.
Surprisingly blunt. :LOL: Maybe he was caught at a bad time. I'm willing to give some benefit of the doubt but if researchers want to use my equipment they need to be responsible. Good communication about the project and addressing bugs quickly are very important.
 

RONNIE

Well-Known Member
USA team member
This is just to tell that, after the THOR Challenge, I intend to allocate most of my resources to help our Folding@home (COVID) team.
 

Vester

Well-Known Member
USA team member
I am running World Community Grid COVID-19 CPU work units on my desktop (AMD Radeon HD 7790) and checking regularly for the anticipated GPU work units. WCG COVID-19 GPU work has been developed and is under IBM review. When the GPU Beta starts, I will put the mining rig to work at WCG instead of MilkyWay@home. As I understand it, the GPU work units will be for AMD GPUs only. From the performance of hardware shown in testing, FP64 performance is rewarded, so bring on the Radeon VIIs to beat my Radeon HD 7990s.
 
Top