Julian Elischer wrote: > this is interesting.. there has been a call to remove "fairness" > as a threading property as it complicates the scheduler. > It has been said by many that they do not consider this as an important > feature and it reduces throughput. IMO both sides are right and fairness should be optional. It will certainly help some workloads and harm others; leaving it as a (default) kernel compile option is fine for now, as long as it's documented how to toggle it (in pthread man page at least).Received on Sun Nov 12 2006 - 18:13:51 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:02 UTC