On Fri, 2006-03-10 at 17:35 +0200, Michael S. Tsirkin wrote:
> Two questions on this
> 1. It is not standard ethernet nor standard IP over Infiniband either, is it?
Correct.
> Is there some documentation on the wire protocol that you use?
No, but the encapsulation is very simple and easy to figure out from
reading the code.
> Is it pathscale specific?
It doesn't have to be.
> 2. Are there practical reasons why you get lower latency and higher
> bandwidth with this than with IPoIB?
The principal reason is that we haven't had time to pay attention to
IPoIB performance yet. The ipath_ether driver was developed before
IPoIB was usable for any real work.
> If there are optimizations, can they not be added to the standard, common
> IP over IB driver?
That's certainly our intention. It's simply a matter of finding the
time for it.
<b
_______________________________________________
openib-general mailing list
[email protected]
http://openib.org/mailman/listinfo/openib-general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
