Joining some other BOINC projects

[networkman]

😀 Tada! Now it's official! 😀


Probably some time later today, I/we (GRRO) will be crossing the one million mark in the Einstein@Home project! 😀 I promised before that once I'd accomplished that, I'd devote some resources to other projects for the TeAm too.

That said, I've fired up the "Rosetta" account again and also joined the "uFluids" and "QMC" projects. So once some WUs get crunched and validated expect to see GRRO showing up on some TA teams. 🙂

Now for the question: Up until now, I've not had to set Resource Shares, but if I leave each project set for say "100", and then join X number of projects on a given machine, each of the projects should get an even slice of the cycles on that machine, correct? If not, then I'm gonna be 😕

 

[amdxborg]

Well a huge congrats to you and GRRO for reaching that amazing milestone!! :beer:😉

To answer your question, I leave all my project shares at 100 and boinc switches projects on the pc every hour or to whatever time you set it in your project preferences.

 

[Wolfsraider]

Originally posted by: amdxborg
Well a huge congrats to you and GRRO for reaching that amazing milestone!! :beer:😉

Absolutely Freakin Awesome Milestone you all have achieved 😛 Go Grand Rapids Radio Observatory!!!!
You are Rocking and Rolling :music:

:thumbsup:

Mike

 

[Fardringle]

If your Resource Share is set to the same number in all projects, then each project will get roughly the same amount of CPU time. If you have a particular project that you want to get more CPU time, either raise its share number, or lower the number on the other projects.

 

[Rattledagger]

Originally posted by: networkman
Probably some time later today, I/we (GRRO) will be crossing the one million mark in the Einstein@Home project! 😀 I promised before that once I'd accomplished that, I'd devote some resources to other projects for the TeAm too.

That said, I've fired up the "Rosetta" account again and also joined the "uFluids" and "QMC" projects. So once some WUs get crunched and validated expect to see GRRO showing up on some TA teams. 🙂

Now for the question: Up until now, I've not had to set Resource Shares, but if I leave each project set for say "100", and then join X number of projects on a given machine, each of the projects should get an even slice of the cycles on that machine, correct? If not, then I'm gonna be 😕

µFluids still doesn't checkpoint, so isn't an option if often re-boots and so on, and some projects can also be heavy on bandwith-usage so can be something to consider for some users, but otherwise it's basically just to choose whatever project(s) you want.

Till one of the "Account Managers" is launched and v5.4.xx is released, the easiest way to join same project to many computers is "attach" on one computer, copy example account_boinc.bakerlab.org_rosetta.xml from boinc-directory on this computer to all the other computers, and re-start BOINC on the other computers.


As for resource-shares, yes, if all projects has the same resource-share they'll get an equal part of the resources. Still, short-term it often doesn't follow this exactly, but long-term it tries to follow whatever user sets.


Also, one last thing to remember is cache-size. If got a permanent connection, my recommended max cache-size is:
0.4*shortest_deadline / N
there N is number of projects.

With example 4 projects and 14-day deadline, max cache-size is 1.4 days.
If example same 4 projects but adds a 5th with 7-days deadline, max cache-size is 0.56 days.

Of course, for dialup-uses or someone only connecting between specific hours or something, the deadline must often be longer than the recommended...


BTW, this also means, when Seti_Enhanced is "soon" released, even someone that only runs SETI@Home should generally not use higher cache-size than 3.5 days, due to fairly short deadline on VHAR. This even majority of wu has over 40 days deadline...

 

[networkman]

Thanks for the info and congrats guys. 🙂

For the time being, I've detached from Rosetta and added Leiden Classical in. I'm sure I'll re-attach to Rosetta once they get this 4.97 client figured out.