Testbed Support for GridPP member institutes [mailto:TB-
> [log in to unmask]] On Behalf Of Alessandra Forti said:
> The problem of Hyperthreading remains. At the moment Manchester and
> QMUL
> are definitely loosing out when the cluster is half full which is
> often.
If the cluster isn't full I'm not entirely sure what's being measured anyway - it means that the site performance is being assessed on the basis of what the VOs happen to submit, which isn't in their control. The only real effect for non-full clusters is that a particular job may finish a bit sooner in some places than others, but is that significant enough for anyone to care? Would any users actively prefer non-hyperthreaded sites?
Stephen
|