Hello! I am running Windows XP SP3 with BOINC 6.4.5 on Dual-core Celeron with 4 simultaneous projects (SETI, Rosetta, LHC and SIMAP). Just know every project has run out of units, except SETI. I have just noticed that BOINC was keeping one finished SETI unit and working on another one (normal). That’s why I updated all projects. All projects but SETI had no units and that’s fine, too. The problem is that BOINC would not request work for the SETI project, too, despite one of the cores is available:

05/01/2009 16:19:41|SETI@home|Sending scheduler request: Requested by user. Requesting 0 seconds of work, reporting 0 completed tasks
05/01/2009 16:19:46|SETI@home|Scheduler request completed: got 0 new tasks

Currently, the load is 53% for the CPU – 1st core is used at about 80% and the second is at about 20%. The only available workunit (SETI) is at 40% progress.
Is this normal? I thought that workunits are distributed through cores, thus forcing BOINC to download unit for every free core available, even when engaged in one project.

BOINC 6.4.5 has a problem with the work fetch module, this bug is known. It can happen that one or more cores go idle although BOINC will not completely run out of work (all cores idle).

It'll fix itself eventually. If you don't like the behaviour and you're not using the CUDA option of BOINC 6.4.5, you can return to BOINC 6.2.19

A fix in the form of a new work fetch policy will come in BOINC 6.8

---

Will wait 6.8. Not that I use the CUDA, but I always like latest versions...
Thanks!!!

Will wait 6.8. Not that I use the CUDA, but I always like latest versions...
Thanks!!!

Not the best strategy (IMO :-), I always stay with the second to last (latest?) version, not only with BOINC, but with the current BOINC development policy (imagine, planning to correct errors (not flaws) 4 (2 production) minor versions in the future!), I'll definitely stay with 5.10.45 for some time.

Gruß,
Gundolf

---

Computer sind nicht alles im Leben. (Kleiner Scherz)