Hmmm - so what happens with IWARP NIC when no buffer is posted on recv q
and a message arrives ?
Or Gerlitz wrote:
On Fri, Apr 4, 2008 at 7:06 PM, Roland Dreier <[EMAIL PROTECTED]> wrote:
- Don't use IB-specific features (atomics, immediate data)
and don't use RNRs as a means for HW based "flow control" mechanism.
The current RDS implementation
does not have a SW based flow control but rather does some sort of
back pressure through SW based congestion
management. I think that to some extent it relies on RNRs which don't
exist under iWARP.
Or.
_______________________________________________
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
_______________________________________________
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
