> Yes, it would be better to do
> perPort/perCompletion/perEvent/perInterrupt. It also helps latency
> not just throughput.

It might even need multiple completions multple Events, multiple interrupt
association for CM mode along with CPU affinity.

thanks
Shirley Ma
_______________________________________________
general mailing list
[email protected]
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general

To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general

Reply via email to