LHC@home

I'm running seti enhanced, einstein and LHC on a few slow machines (2 linux, 1 XP), and I'm wondering whether anyone has any ideas how best to deal with LHC's spotty work availability. I want to do as much LHC as possible, and do the other projects the rest of the time, when LHC is unavailable.

With roughly equal resource shares, LHC's work often runs out before my slow machines have finished crunching the WUs it got at the beginning. I thought the debt accumulated to LHC during dry periods would have the effect of prioritizing it when work became available, but boinc kept crunching the other units in round-robin fashion, until they finished, so I didn't actually do much LHC before work ran out

I'm currently trying a massively larger resource share (1000) for LHC, and smaller ones for s@h and e@h (100 each), and that seems to be doing the business now, while there is work from LHC. It is prioritising as I expect.

But what about when LHC goes quiet again? If I don't change anything, I think I'll develop massive debts to LHC, which will interfere with the proper running of the other projects. FWIW I'm on a permanent connection, with a small cache size (0.04 days).

What should I do? I want a solution that doesn't involve me having to suspend or NNW projects. Or is intermittent work impossible to handle properly in boinc as it stands?

P

---

I'm running seti enhanced, einstein and LHC on a few slow machines (2 linux, 1 XP), and I'm wondering whether anyone has any ideas how best to deal with LHC's spotty work availability. I want to do as much LHC as possible, and do the other projects the rest of the time, when LHC is unavailable.

With roughly equal resource shares, LHC's work often runs out before my slow machines have finished crunching the WUs it got at the beginning. I thought the debt accumulated to LHC during dry periods would have the effect of prioritizing it when work became available, but boinc kept crunching the other units in round-robin fashion, until they finished, so I didn't actually do much LHC before work ran out

I'm currently trying a massively larger resource share (1000) for LHC, and smaller ones for s@h and e@h (100 each), and that seems to be doing the business now, while there is work from LHC. It is prioritising as I expect.

But what about when LHC goes quiet again? If I don't change anything, I think I'll develop massive debts to LHC, which will interfere with the proper running of the other projects. FWIW I'm on a permanent connection, with a small cache size (0.04 days).

What should I do? I want a solution that doesn't involve me having to suspend or NNW projects. Or is intermittent work impossible to handle properly in boinc as it stands?

P

I think your best bet would be to trust the BOINC client to the right "thing" when the time comes. BOINC was designed to allow for exactly that which you want: use LHC as a primary project.

To ensure that LHC remains the primary, set you resource share to some really high number such as 500000 and the other "backup" projects to a resource share of 1. This will ensure that when LHC has work, you will get it. When LHC does not (as much of the time this is the case) your other projects will kick in and start downloading WU's.

It is true that LHC will continue to add debt but when new WU's become available LHC will go back to normal crunching splitting the time with the other projects (per your resource share) until they are finished or until round robin robin kicks in. BOINC will finish any and all WU's before the deadline as it was designed to do. Just trust the client to do what it is supposed to do :)

I hope this helps some. I don't claim to be an expert of any of this but this is how I understand it. Where's River when you need him...

Tim

---

Check out my blog Singularity Now

sslickerson share suggestion (500000:1:1) is a good one. Einstein has a 2 week due date and Seti now has even longer due dates on most WU. LHC should easily get crunching priority and Boinc will only bring your low share projects into the mix when necessary to meet their deadlines. A cache of .2 days would ensure that you have a couple of LHC WU on hand while they are available. The large resource share and the long term debt will see to it that you will not get new work from the other projects while LHC still has work available.
As long as you don't try to micromanage, Boinc will put all the emphasis on doing LHC.

Happy crunching.

---

Questions? Answers are in the BOINC Wiki.

Boinc V6.10.56 Recommended
WinXP C2D 2.1G 3GB

Thanks for the information. I was wondering about this, too. I'll try to make some modifications to my settings before the next round of work becomes available and see if it helps.

---

Thanks for your help. The biased resource shares seem to have worked well this time.

---