On 5/23/2018 3:44 PM, Jason Gunthorpe wrote:
On Wed, May 23, 2018 at 02:08:19PM -0400, Dennis Dalessandro wrote:
On 5/23/2018 12:48 PM, Parav Pandit wrote:
- log_pmtu = ah->log_pmtu;
+ log_pmtu = ilog2(ah->pmtu);
I don’t know if UD is widely used in IPoIB or not.
But won't you have any performance impact by doing ilog2 on every work request?
Just curious.
It might just find more use with a larger MTU that allows better
performance.
At least on mlx5 ipoib and netdev ip o ethernet performance are
basically the same. The way to get good performance is to implement
the IP offloads.
As I've said before, if OPA cares about good ipoib performance then
the hfi driver needs to implement the rdma_netdev interface for ipoib
as well as for opa_vnic.
But, increasing the allowed MTU is a reasonablish idea too..
Jason
Actually, we've discussed this a bit more internally, let's go ahead and
drop this patch, for now at least.
Thanks
-Denny
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
