Testbed Support for GridPP member institutes [mailto:TB-
> [log in to unmask]] On Behalf Of Alex Martin said:
> All the current accounting assumes the performance of
> the CPU performance per task is independent of the number of tasks running in
> total (which has probably never been the case). So if there are less than
> N jobs running they will tend to run faster than accounted for.
> Hyperthreading makes this much worst.
The theory is that sites should be running close to full most of the time. And you might argue that a half-empty site should be "charging" less anyway, similar to a flight with lots of empty seats ...
Stephen
|