On 02/23/2013 09:12 AM, Louis Letourneau wrote: > Admins of mammouth installed a new version of mpi > mvapich2_gcc64/1.9a2 > > Recompiled and tried it. After 11hrs it crashed with the message below. > Not sure if this is from ray or the mpi implementation. > > I'll restart with my previous mvapich2 1.6 ofed. > Just wondering if this means something to you. > > Thanks > Louis > > > Got unknown event 17 ... continuing ... > (Many lines like this) > > rank 800 is creating seeds [3200001/6842888]ank 800: assembler memory usage: > 819912 KiB > [0->495] send desc error, wc_opcode=1 > [0->495] wc.status=12, wc.wr_id=0x2356610, wc.opcode=1, vbuf->phead->type=0 = > MPIDI_CH3_PKT_EAGER_SEND > [504] Abort: [] Got completion with error 12, vendor code=0x81, dest rank=495 > at line 586 in file ibv_channel_manager.c >
Sounds like a problem with MVAPICH2 (or possibly the Infiniband fabric). ------------------------------------------------------------------------------ Everyone hates slow websites. So do we. Make your web apps faster with AppDynamics Download AppDynamics Lite for free today: http://p.sf.net/sfu/appdyn_d2d_feb _______________________________________________ Denovoassembler-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/denovoassembler-users
