Re: [OMPI devel] large virtual memory consumption on smp nodes and gridengine problems

Mon, 11 Jun 2007 08:05:31 -0400 

In addition to what Ralph said, I have the following random comments:
1. You'll have to ask on the GE mailing lists about the GE issues (2gb vs. 2000mb, etc.); I doubt we'll be of much help here on this list.
2. Do you have a very large SMP machine (i.e., 16 cores or more)? More specifically, how many MPI processes do you plan to run at once on a host?
3. Unrelated to the SMP issue, I see that you are using the InfiniBand Mellanox VAPI interface (mvapi BTL). Is there any chance that you can upgrade to the newer OpenFabrics kernel/user stack? All the IB vendors support it for their HPC customers. FWIW: all Open MPI InfiniBand work is being done in support of OpenFabrics; the "mvapi" BTL is only maintained for backward compatibility and has had no new work done on it in at least a year. See http://www.open- mpi.org/faq/?category=openfabrics#vapi-support.
4. Note that depending on your application (e.g., if it primarily sends large messages), it *may* be faster to use the DMA engine in your IB interface and not use Open MPI's shared memory interface. But there are a lot of factors involved here, such as the size of your typical messages, how many processes you run per host (i.e., I'm assuming you have one HCA that would need to service all the processes), etc. 


On Jun 10, 2007, at 6:04 PM, Ralph Castain wrote:


Hi Markus

There are two MCA params that can help you, I believe:

1. You to set the maximum size of the shared memory file with

-mca mpool_sm_max_size xxx

where xxx is the maximum memory file you want, expressed in bytes. The
default value I see is 512MBytes.

2. You can set the size/peer of the file, again in bytes:

-mca mpool_sm_per_peer_size xxx
This will allocate a file that is xxx * num_procs_on_the_node on each node, 
up to the maximum file size (either the 512MB default or whatever you
specified using the previous param). This defaults to 32MBytes/proc.
I see that there is also a minimum (total, not per-proc) file size that defaults to 128MBytes. If that is still too large, you can adjust it using 

-mca mpool_sm_min_size yyy


Hope that helps
Ralph
On 6/10/07 2:55 PM, "Markus Daene" <markus.da...@physik.uni- halle.de> wrote: 


Dear all,

I hope I am in the correct mailing list with my problem.
I try to run openmpi with the gridengine(6.0u10, 6.1). Therefore I
compiled openmpi (1.2.2),
which has the gridengine support included, I have checked it with ompi_info. 
In principle, openmpi runs well.
The gridengine is configured such that the user has to specify the
memory consumption
via the h_vmem option. Then I noticed that with a larger number of
processes the job
is killed by the gridengine because of taking too much memory.
To take a closer look on that, I wrote a small and simple (Fortran) MPI 
program which has just a MPI_Init
and a (static) array, in my case of 50MB, then the programm goes into a 
(infinite) loop, because it
takes some time until the gridengine reports the maxvmem.
I found, that if the processes run all on different nodes, there is only 
a offset per process, at least
a linear scaling. But it becomes worse when the jobs run on one node.
There it seems to be a quadratic
scaling with the offset, in my case about 30M. I made a list of the
virtual memory reported by the
gridengine, I was running on a 16 processor node:

#N proc    virt. Mem[MB]
1          182
2          468
3          825
4          1065
5          1001
6          1378
7          1817
8          2303
12         4927
16         8559

the pure program should need N*50MB, for 16 it is only 800M, but it
takes 10 times more, >7GB!!!
Of course, the gridengine will kills the job is this overtaking is not 
taken into accout,
because of too much virtual memory consumption. The momory consumtion is 
not related to the grid engine,
it is the same if I run from the command line.
I guess it might be related to the 'sm' component of the btl.
Is it possible to avoid the quadratic scaling?
Of course I could use the vapi/tcp component only like
mpirun --mca btl mvapi  -np 16 ./my_test_program
in this case the virtual memory is fine, but it will not be what one
wants on a smp node.


then it becomes ever worse:
openmpi nicely report the (max./act.) used virtual memory to the grid
engine as sum of all processes.
This value is the compared with the one the user has specified with the 
h_vmem option, but the
gridengine takes this value per process for the allocation of the job
(works) and does not multiply
this with the number of processes. Maybe one should report this to the 
gridenging mailing list, but it
could be related as well for the openmpi interface.

The last thing I noticed:
It seems that if the v_mem option for gridengine jobs is specified like 
'2.0G' my test job was
immedialtely killed; but when I specify '2000M' (which is obviously
less) it work. The gridengine
puts the job allways on the correct node as requested, but I think there 
is might be a problem in
the openmpi interface.


It would be nice if someone could give some hints how to avoid the
quadratic scaling or maybe to think
if this is really neccessary in openmpi.


Thanks.
Markus Daene




my compiling options:
./configure --prefix=/not_important --enable-static
--with-f90-size=medium --with-f90-max-array-dim=7  --with-mpi-para
m-check=always --enable-cxx-exceptions --with-mvapi
--enable-mca-no-build=btl-tcp

ompi_info output:
                Open MPI: 1.2.2
   Open MPI SVN revision: r14613
                Open RTE: 1.2.2
   Open RTE SVN revision: r14613
                    OPAL: 1.2.2
       OPAL SVN revision: r14613
                  Prefix: /usrurz/openmpi/1.2.2/pathscale_3.0
 Configured architecture: x86_64-unknown-linux-gnu
           Configured by: root
           Configured on: Mon Jun  4 16:04:38 CEST 2007
          Configure host: GE1N01
                Built by: root
                Built on: Mon Jun  4 16:09:37 CEST 2007
              Built host: GE1N01
              C bindings: yes
            C++ bindings: yes
      Fortran77 bindings: yes (all)
      Fortran90 bindings: yes
 Fortran90 bindings size: small
              C compiler: pathcc
     C compiler absolute: /usrurz/pathscale/bin/pathcc
            C++ compiler: pathCC
   C++ compiler absolute: /usrurz/pathscale/bin/pathCC
      Fortran77 compiler: pathf90
  Fortran77 compiler abs: /usrurz/pathscale/bin/pathf90
      Fortran90 compiler: pathf90
  Fortran90 compiler abs: /usrurz/pathscale/bin/pathf90
             C profiling: yes
           C++ profiling: yes
     Fortran77 profiling: yes
     Fortran90 profiling: yes
          C++ exceptions: yes
          Thread support: posix (mpi: no, progress: no)
  Internal debug support: no
     MPI parameter check: always
Memory profiling support: no
Memory debugging support: no
         libltdl support: yes
   Heterogeneous support: yes
 mpirun default --prefix: no
MCA backtrace: execinfo (MCA v1.0, API v1.0, Component v1.2.2) MCA memory: ptmalloc2 (MCA v1.0, API v1.0, Component v1.2.2) MCA paffinity: linux (MCA v1.0, API v1.0, Component v1.2.2) MCA maffinity: first_use (MCA v1.0, API v1.0, Component v1.2.2) MCA maffinity: libnuma (MCA v1.0, API v1.0, Component v1.2.2) MCA timer: linux (MCA v1.0, API v1.0, Component v1.2.2) 
         MCA installdirs: env (MCA v1.0, API v1.0, Component v1.2.2)
MCA installdirs: config (MCA v1.0, API v1.0, Component v1.2.2) 
           MCA allocator: basic (MCA v1.0, API v1.0, Component v1.0)
           MCA allocator: bucket (MCA v1.0, API v1.0, Component v1.0)
MCA coll: basic (MCA v1.0, API v1.0, Component v1.2.2) 
                MCA coll: self (MCA v1.0, API v1.0, Component v1.2.2)
                MCA coll: sm (MCA v1.0, API v1.0, Component v1.2.2)
MCA coll: tuned (MCA v1.0, API v1.0, Component v1.2.2) MCA io: romio (MCA v1.0, API v1.0, Component v1.2.2) 
               MCA mpool: rdma (MCA v1.0, API v1.0, Component v1.2.2)
               MCA mpool: sm (MCA v1.0, API v1.0, Component v1.2.2)
                 MCA pml: cm (MCA v1.0, API v1.0, Component v1.2.2)
                 MCA pml: ob1 (MCA v1.0, API v1.0, Component v1.2.2)
                 MCA bml: r2 (MCA v1.0, API v1.0, Component v1.2.2)
              MCA rcache: vma (MCA v1.0, API v1.0, Component v1.2.2)
MCA btl: self (MCA v1.0, API v1.0.1, Component v1.2.2) 
                 MCA btl: sm (MCA v1.0, API v1.0.1, Component v1.2.2)
MCA btl: mvapi (MCA v1.0, API v1.0.1, Component v1.2.2) MCA topo: unity (MCA v1.0, API v1.0, Component v1.2.2) MCA osc: pt2pt (MCA v1.0, API v1.0, Component v1.2.2) 
              MCA errmgr: hnp (MCA v1.0, API v1.3, Component v1.2.2)
MCA errmgr: orted (MCA v1.0, API v1.3, Component v1.2.2) MCA errmgr: proxy (MCA v1.0, API v1.3, Component v1.2.2) 
                 MCA gpr: null (MCA v1.0, API v1.0, Component v1.2.2)
MCA gpr: proxy (MCA v1.0, API v1.0, Component v1.2.2) MCA gpr: replica (MCA v1.0, API v1.0, Component v1.2.2) MCA iof: proxy (MCA v1.0, API v1.0, Component v1.2.2) 
                 MCA iof: svc (MCA v1.0, API v1.0, Component v1.2.2)
MCA ns: proxy (MCA v1.0, API v2.0, Component v1.2.2) MCA ns: replica (MCA v1.0, API v2.0, Component v1.2.2) 
                 MCA oob: tcp (MCA v1.0, API v1.0, Component v1.0)
MCA ras: dash_host (MCA v1.0, API v1.3, Component v1.2.2) MCA ras: localhost (MCA v1.0, API v1.3, Component v1.2.2) MCA ras: gridengine (MCA v1.0, API v1.3, Component v1.2.2) MCA ras: slurm (MCA v1.0, API v1.3, Component v1.2.2) MCA rds: hostfile (MCA v1.0, API v1.3, Component v1.2.2) MCA rds: proxy (MCA v1.0, API v1.3, Component v1.2.2) MCA rds: resfile (MCA v1.0, API v1.3, Component v1.2.2) MCA rmaps: round_robin (MCA v1.0, API v1.3, Component v1.2.2) MCA rmgr: proxy (MCA v1.0, API v2.0, Component v1.2.2) 
                MCA rmgr: urm (MCA v1.0, API v2.0, Component v1.2.2)
                 MCA rml: oob (MCA v1.0, API v1.0, Component v1.2.2)
MCA pls: proxy (MCA v1.0, API v1.3, Component v1.2.2) MCA pls: gridengine (MCA v1.0, API v1.3, Component v1.2.2) 
                 MCA pls: rsh (MCA v1.0, API v1.3, Component v1.2.2)
MCA pls: slurm (MCA v1.0, API v1.3, Component v1.2.2) 
                 MCA sds: env (MCA v1.0, API v1.0, Component v1.2.2)
                 MCA sds: seed (MCA v1.0, API v1.0, Component v1.2.2)
MCA sds: singleton (MCA v1.0, API v1.0, Component v1.2.2) 
                 MCA sds: pipe (MCA v1.0, API v1.0, Component v1.2.2)
MCA sds: slurm (MCA v1.0, API v1.0, Component v1.2.2) 

----------------------------------------------------------
Markus Daene
Martin Luther University Halle-Wittenberg
Naturwissenschaftliche Fakultaet II
Institute of Physics
Von Seckendorff-Platz 1 (room 1.28)
06120 Halle
Germany

_______________________________________________
devel mailing list
de...@open-mpi.org
http://www.open-mpi.org/mailman/listinfo.cgi/devel



_______________________________________________
devel mailing list
de...@open-mpi.org
http://www.open-mpi.org/mailman/listinfo.cgi/devel



--
Jeff Squyres
Cisco Systems

Reply via email to