Hi Scott

I have a suggestion for said title.

Last year nano-hive@home needed fast turn around with a 1 day deadline.What they did to not restrict hosts was to allow only x amount of tasks per cpu...lets say 3...you were not given another result until you only had 2 in queue for 1-cpu,5 for 2-cpu's,11 for 4-cpu's etc..with still a max daily amount per cpu of lets say 50.

That way you don't throttle the faster cpu's and get a huge increase in throughput.Also everyone is able to get work with no hogging.

I'm sure Brian(aka Crunch1) at nano would be happy to help you in the how if you are interested.

Thanks JR

---

Scott.

If you're interested in this, I can put you in touch with Brian. I'm one of the mods over there.

---

Kathryn :o)
The BOINC FAQ Service
The Unofficial BOINC Wiki
The Trac System
More BOINC information than you can shake a stick of RAM at.

I may look into this at a later time, but there are other technical parts of the project that need attention for the time being.

Thank you for the suggestion, though!

---

Scott Kruger
Project Administrator, Cosmology@Home

Scott- After 4 days of crunching cosmo I realized in essence you are already using my suggestion....ie: I get 50 units loaded on a host at a time and are allowed new ones only when I report...seems to work ok with the 500 limit even with the top machines :) So the per host rather than per cpu works ok. Hats off to the current process! I spoke too soon. Maybe increased load will take you down to my example. JR

---

If it ain't broke, don't fix it, right?

I'm keep your suggestion in mind though, in case a change needs to be made.

---

Scott Kruger
Project Administrator, Cosmology@Home

I think, the work distribution works fine. I got all 20 minutes deffering time 2 WU´s ready and so I can catch 2 new WU´s. The problem of the 500 daily limit I don´t have with my little host. ;)