<html><body>
<p><tt>rdma-developers-admin@lists.sourceforge.net wrote on 05/27/2005 10:25:36 AM:<br>
<br>
> Mike,<br>
> I am not sure I do understand what your were trying<br>
> to communicate. Let me try and decode this. My basic<br>
> point was to respond to Venkata's response about<br>
> complete offload without any interaction with the host<br>
> system. I disagree with that in its totality as I<br>
> think there are dependencies and that needs to be<br>
> specified in a formal manner, i.e. in the specs, so<br>
> that we do not have multiple proprietary interafaces<br>
> which change and application users have to change this<br>
> consumption based on individual implementations. <br>
> And you just brought up the reasons why what I was<br>
> saying seemed to be justfied. Did I read that<br>
> correctly?<br>
> <br>
> Thanks<br>
> SG<br>
> </tt><br>
<br>
Hello Sukanta,<br>
<br>
<tt>Yes, there will be interactions with the host system for an RNIC</tt><br>
<tt>but through an interface something similar to RNIC-PI.</tt><br>
<br>
The clarification that I was trying to make is that from an RNIC perspective, it <br>
doesn't have to expose any ToE interface other than the fact that adapters might be providing<br>
multiple services including ToE, RDMA, iSCSI but that's outside the scope of an<br>
RNIC interface. <br>
<br>
<tt>Thanks</tt><br>
<tt>Venkat</tt></body></html>