[mvapich-discuss] process limits
Dhabaleswar Panda
panda at cse.ohio-state.edu
Tue Aug 28 09:55:55 EDT 2007
Hi Mark,
> Is there an effective or hard limit on the number of MVAPICH
> processes that can be run on a single node?
>
> Given N cpus, each having M cores, on a single node, I've been told
> that one can not run more than N*M MVAPICH processes on a single
> node. In fact, I observe that if I try to even approach this number
> with "-np 16" (for a node with N=8 and M=4), I observe a "unable to
> find child nnnn!" or "Child died" message. Is this a configuration
> problem with this system or somehow an expected behavior?
Not sure why you are getting this error. We have run applications with
N*M MVAPICH processes with N CPUs each having M cores (N=2, M=4; N=4,
M=2).
Can you provide more details on the system configuration, the MVAPICH
version you are running and the interface you are using (Gen2, uDAPL,
shared memory, etc.).
> More pointedly, should oversubscription of cores, np > N*M, on a
> single node work in MVAPICH? How about in MVAPICH2?
Yes, oversubscription should work for both MVAPICH and
MVAPICH2. However, for oversubscription to work, you need to use the
`blocking' mode of operation, not `polling' mode (which is the normal
mode of operation to get the best performance). This is typical for
any MPI design. MVAPICH 0.9.9 supports `blocking' mode. The latest
version of MVAPICH2 (1.0-beta, soon we will have the final version)
also supports `blocking' mode. Please take a look at the respective
user guides to see how to use the `blocking' mode.
Thanks,
DK
> regards,
> --
> ***********************************
> >> Mark J. Potts, PhD
> >>
> >> HPC Applications Inc.
> >> phone: 410-992-8360 Bus
> >> 410-313-9318 Home
> >> 443-418-4375 Cell
> >> email: potts at hpcapplications.com
> >> potts at excray.com
> ***********************************
> _______________________________________________
> mvapich-discuss mailing list
> mvapich-discuss at cse.ohio-state.edu
> http://mail.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss
>
More information about the mvapich-discuss
mailing list