>Yes, the memory seems increasing in the task manager when the test is >running.
I do see a memory leak on the client side, when running over mlx. I don't see any leak when running over mthca, which is what I usually test over. I'll try to trace this down. Also, with the timing changes resulting from the pkey caching, I see a race on my mlx system handing a DREQ before an RTU has been received. This can result in NDAccept failing, so I'll work on a fix for that as well. - Sean _______________________________________________ ofw mailing list [email protected] http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ofw
