My immediate thought is CPDN, where I only want one WU at the time, as they usually hinder each other. At least that was the case for the most effective for my machine, which are currently not available. I don't know what was (or may still be) the bottleneck, I just know that 2 at the same time slowed considerably down compared to one only.
I'd like it as well for multi-core apps, to guarantee that they leave a certain number of cores free for other projects. > OK, a "no more than nCPUs" is a bit different than "exactly nCPUs", but > would be better served if we could figure it out automatically somehow. > > What resource(s) are required that are causing the problem? Is there some > way of detecting this conflict? If we can answer both, we would be much > better off doing it automatically. > > jm7 > _______________________________________________ boinc_dev mailing list [email protected] http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev To unsubscribe, visit the above URL and (near bottom of page) enter your email address.
