<html><body>
<p><tt>> > Regarding the memory issue w/o SRQ, do you think there is a way to use low<br>
> > watermark to release prepost buffer in large connections?<br>
> <br>
> Maybe with UC - with RC you'll get RNR and connection'll get closed before you<br>
> have time to handle the low watermark. So sure, might be an interesting idea,<br>
> but isn't low watermark a SRQ feature?<br>
> <br>
> > I think most of the prepost buffers are empty in that case becauseof the BW.<br>
> <br>
> I don't really get the argument.<br>
> <br>
> -- <br>
> MST<br>
</tt><br>
<tt> That's just some random idea. :) Some other ideas like to share RQ buffer based on source-destination address, per CPU RQ buffer ...which might hurt performance too much?</tt><br>
<br>
<tt> It might be too complicated to have UD/RC mode coexisted?</tt><br>
<br>
<tt> Maybe it's better to set up a small RQ size for now, and later when high watermark patch is available we can use it to address RQ overrun?</tt><br>
<br>
<tt>Thanks</tt><br>
<tt>Shirley</tt></body></html>