Charles Elliott wrote: > About 5%. Been doing it for over a year using a program written in C++ and > operating as a service under Windows.
Sorry, but I think we'll need that confirming by some means for such a high claimed boost. I don't how well Windows does it's scheduling, but my observation on Linux is that for a 'quiet' system, a process will stay on a single CPU core for very long periods of time. Hence, there should be no need for 'forcing' CPU affinity. Indeed, restricting the CPU affinity could well show a reduced performance in some circumstances. As always, the truth is in any real world measurements! Happy crunchin', Martin _______________________________________________ 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.
