Jim Phillips

*Original date: 2017-09-14 15:45:19* --- The bug does not affect the MPI layer on Blue Waters. Still need to test verbs.

*Original date: 2017-09-15 14:56:44* --- verbs layer does not appear to be affected.

*Original date: 2017-09-15 21:33:32* --- Correct, as far as I know this is a GNI issue. I've only tested on Blue Waters. It may or may not affect Titan, Eos,...

*Original date: 2017-10-01 10:10:34* --- I can confirm that the bug also affects Cori (XC40), so I would assume all XC/XE/XK machines.

*Original date: 2017-10-03 19:14:22* --- No and no. I've been using 4 nodes per replica, non-smp. The effect starts to be visible above noise at 16 replicas, stands out at...

*Original date: 2017-10-03 19:32:24* --- All of the replicas are uniformly slow. There is no inter-replica interaction. I haven't looked at large node counts with small replica counts.

*Original date: 2017-10-04 22:13:28* --- From some basic profiling it appears that the amount of time spent in alloc_mempool_block (but not the number of calls) increases dramatically as the number...

*Original date: 2017-10-06 02:20:41* --- For 16 nodes: aprun -n 496 -r 1 -N 31 -d 1 /u/sciteam/jphillip/NAMD_LATEST_CRAY-XE-ugni-BlueWaters/namd2 +pemap 0-30 +useDynamicSmsg +replicas 4 +stdout output/%d/test22.%d.log /u/sciteam/jphillip/apoa1/apoa1.namd From a 20-node run:...

*Original date: 2017-10-06 02:59:27* --- Sorry, dynamic and static SMSG have indistinguishable performance at large replica counts, although the final WallClock output is actually longer for dynamic SMSG than for...

Thanks for the reply. I noticed it while reading the code to investigate an unrelated issue. I haven't observed any consequences, and I can't guess if fixing it would break...