From: Tim.Ho_at_csiro.au
Date: Wed Apr 05 2017 - 01:53:26 CDT
Hi,
I built NAMD 2.12 on our cluster and tried to run apoa1 (which runs fine with v2.11, by the way) and got the below error:
FATAL ERROR: PmePencilXYMap::populateInitial, multiple pencils on a pe?
FATAL ERROR: See http://www.ks.uiuc.edu/Research/namd/bugreport.html
------------- Processor 0 Exiting: Called CmiAbort ------------
Reason: FATAL ERROR: PmePencilXYMap::populateInitial, multiple pencils on a pe?
FATAL ERROR: See http://www.ks.uiuc.edu/Research/namd/bugreport.html
[0] Stack Traceback:
[0:0] _Z8NAMD_bugPKc+0xa8 [0x725e38]
[0:1] _ZN14PmePencilXYMap15populateInitialEiR14CkArrayOptionsPvP8CkArrMgr+0x10a [0x110444a]
[0:2] _ZN7CkArrayC1ER14CkArrayOptionsR19CkMarshalledMessage10_ckGroupID+0xbfc [0x13aa1fc]
[0:3] _ZN15CkIndex_CkArray23_call_CkArray_marshall1EPvS0_+0x158 [0x13a94b8]
[0:4] CkCreateLocalGroup+0x9e4 [0x1348dc4]
[0:5] CkCreateGroup+0x16c [0x1349e7c]
[0:6] _ZN14CProxy_CkArray5ckNewERK14CkArrayOptionsRK19CkMarshalledMessageRK10_ckGroupIDPK14CkEntryOptions+0x180 [0x13a0f00]
[0:7] _ZN16CProxy_ArrayBase13ckCreateArrayEP14CkArrayMessageiRK14CkArrayOptions+0x595 [0x13a74c5]
[0:8] _ZN22CProxy_CudaPmePencilXY5ckNewERK14CkArrayOptions+0x68 [0xce5138]
[0:9] _ZN17ComputePmeCUDAMgr10initializeEP7CkQdMsg+0x818 [0xbf2ff8]
[0:10] [0x134bdb8]
[0:11] _Z15_processHandlerPvP11CkCoreState+0x208 [0x134ac98]
[0:12] CsdScheduler+0x480 [0x14cf0c0]
[0:13] _ZN9ScriptTcl3runEv+0x2b9 [0x115c829]
[0:14] main+0x4e1 [0x7411b1]
[0:15] __libc_start_main+0xf5 [0x7ffff6352b25]
[0:16] [0x6340ab]
I am using OpenMPI 1.10.6 and Intel compiler 2017 update 2 on SLES. There are 4 GPUs on each node.
I followed the build instructions on http://www.ks.uiuc.edu/Development/Download/download.cgi.
The command I ran was "mpirun -np 4 namd2 +idlepoll apoal1.namd".
Note that using one worker apoa1 runs fine.
Any idea what is causing the fatal error?
Cheers,
Tim
This archive was generated by hypermail 2.1.6 : Sun Dec 31 2017 - 23:21:11 CST