[mvapich-discuss] error in MPI_Igatherv

[Konz, Jeffrey (SSA Solution Centers)]

Is there any update on a work around to this error?

Thanks,
-Jeff
  

> 
> Today's Topics:
> 
>    1. error in MPI_Igatherv (Konz, Jeffrey (SSA Solution Centers))
> 
> Message: 1
> Date: Wed, 28 May 2014 20:25:04 +0000
> From: "Konz, Jeffrey (SSA Solution Centers)" <jeffrey.konz at hp.com>
> To: "mvapich-discuss at cse.ohio-state.edu"
> 	<mvapich-discuss at cse.ohio-state.edu>
> Subject: [mvapich-discuss] error in MPI_Igatherv
> Message-ID:
> 	<77486CDD5CFC9449BEE3C555886D8A908B7DBE37 at G4W3213.americ
> as.hpqcorp.net>
> 
> Content-Type: text/plain; charset="us-ascii"
> 
> Running a Fortran90 code I get this error:
> 
> Fatal error in PMPI_Igatherv:
> Other MPI error, error stack:
> PMPI_Igatherv(324)........:  MPI_Igatherv(sendbuf=0x2aaab8116aa0,
> sendcount=5184, sendtype =MPI_REAL, recvbuf=0x2aaab8400010, recvcounts=0x2aaabef75d30,
> displs=0x2aaabef67c00, recvtype=MPI_REAL, root=55, comm=MPI_COMM_WORLD,
> request=0x2aaabef7c53c)
> 
> PMPI_Igatherv(310)........:
> MPIR_Igatherv_impl(140)...:
> MPIR_Igatherv(69).........:
> MPID_Sched_recv(599)......:
> MPIDU_Sched_add_entry(425): Out of memory
> 
> This error occurs when running on 120 nodes (2400 cores). Job runs fine on 90
> nodes (1800 cores).
> 
> This is with the just released version 2.0-rc2.