Rob > Thanks for the graphs. I was surprised to see that fewer than 8 cores per node resulted in slightly *worse* performance for collective I/O. <
me too. My theory is that as BGQ has IO forwarding, for the small node counts (=small data transfers too) the forwarding directly is more effective than collecting first and then forwarding. I am also a little puzzled the shape of the graphs later as independent drops and collective overtakes it. I presume the effect of latency on many writes to IONs is causing performance to drop whilst the collective mode avoids some of this. This page of graphs is one of many for different file system configs and they all show the same pattern to greater or lesser degree. I shall do more tests I expect until I am happy that I fully understand what's going on. JB _______________________________________________ Hdf-forum is for HDF software users discussion. [email protected] http://mail.lists.hdfgroup.org/mailman/listinfo/hdf-forum_lists.hdfgroup.org
