[ofa-general] infiniband multicast (libibverbs)

[Devesh Sharma]

are you taking care that ibv_poll_cq is not a blocking call, I mean you are
not considering it as blocking call and just going ahead with the sequence
number check?

On 11/5/08, Kelly Burkhart <kelly at tradebotsystems.com> wrote:
>
>
>
> > -----Original Message-----
> > From: Roland Dreier [mailto:rdreier at cisco.com]
> >
> >  > If I specify more than about 6500 messages/sec, I skip
> > some sequences
> >  > and receive others multiple times.  I always receive the
> > same number of
> >  > messages the client sent.  It appears as though all of the
> > messages come
> >  > through, but I'm missing some and reading others twice.
> >
> > Sounds like a bug in your code -- I don't know why you would see
> > duplicate messages unless you are somehow processing the same receive
> > buffer twice or something like that.
>
> I am (or was) processing the same buffer over and over.  I ripped
> from ib_send_lat which does the same thing.  The difference is
> send_lat waits for a reply before sending a second message.  I'm
> sending rapidly without waiting for a reply.  The surprising thing
> to me was that my recv buffer received data ahead of me waiting on
> the cq.
>
> I modified my code to read into a circular list of buffers which
> appears to have solved the problem at the cost of more memory usage.
>
> Thanks,
>
> -K
> _______________________________________________
> general mailing list
> general at lists.openfabrics.org
> http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general
>
> To unsubscribe, please visit
> http://openib.org/mailman/listinfo/openib-general
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openfabrics.org/pipermail/general/attachments/20081105/a8d92105/attachment.html>