[MPICH] MPICH2 v1.02 in heterogeneous enviroment

[Rajeev Thakur]

Yes it's definitely planned, but can't give a firm date for when it will
appear in the implementation.
 
Rajeev
 


  _____  

From: Kenneth Rempe [mailto:kenneth.rempe at studsvikscandpower.com] 
Sent: Tuesday, March 07, 2006 2:16 PM
To: Rajeev Thakur
Cc: mpich-discuss at mcs.anl.gov
Subject: Re: [MPICH] MPICH2 v1.02 in heterogeneous enviroment


Thanks for the response. I found the following in the MPICH2_1.0.3 release
notes:

- The CH3 device does not presently support heterogeneous communication.
That
  is to say that the processes involved in a job must use the same basic
type
  sizes and format.  The sizes and format are typically determined by the
  processor architecture, although it may also be influenced by compiler
  options.  This device does support the use of different executables (e.g.,
  multiple-program-multiple-data, or MPMD, programming).


Which is just as you stated. Do you know if true heterogeneous support is
planned?

Thanks again.

Ken



Rajeev Thakur wrote:


MPICH2 doesn't yet work in heterogeneous environments. It will work in

heterogeneous environments that really are homogoneous, ie, no differences

in byte ordering, sizes of datatypes, etc. 



Rajeev

 



  

-----Original Message-----

From: owner-mpich-discuss at mcs.anl.gov 

[mailto:owner-mpich-discuss at mcs.anl.gov] On Behalf Of Kenneth Rempe

Sent: Monday, March 06, 2006 8:13 AM

To: mpich-discuss at mcs.anl.gov

Subject: [MPICH] MPICH2 v1.02 in heterogeneous enviroment





I'm using MPICH2 v1.02 with IBM AIX 5.2 and Red Hat Linux. My program 

works fine

if I just run on Linux machines or just IBM machines but 

fails with the 

following message when

trying to run on both IBM and Linux at the same time.



Fatal error in MPI_Comm_spawn: Internal MPI error!, error stack:

MPI_Comm_spawn(128): 

MPI_Comm_spawn(cmd="/home/ken/DUKE/Formosa-link/New/FPSIM3/Lin

ux/fpsim3.exe", 

argv=2005bc88, maxprocs=1, info=0x9c000000, root=0, MPI_COMM_WORLD, 

intercomm=2005b858, errors=2002fff0) failed

MPID_Comm_spawn_multiple(52):

MPIDI_CH3_Comm_spawn_multiple(212):

MPIDI_CH3_Comm_accept(102):

MPIDI_CH3_Progress_wait(209): an error occurred while 

handling an event 

returned by MPIDU_Sock_Wait()

MPIDI_CH3I_Progress_handle_sock_event(886): [ch3:sock] 

received packet 

of unknown type (385875968)

% rank 0 in job 1  ibm6e1_34162   caused collective abort of all ranks

  exit status of rank 0: killed by signal 9





Is it possible to use MPICH2 in a heterogeneous environment?



Thanks.





    



  



-- 

*** NOTE: soa.com will no longer be used after February 1, 2006, please
change your address book ***



Kenneth R. Rempe                * email:
kenneth.rempe at studsvikscandpower.com

Studsvik Scandpower, Inc.       * 

504 Shoup Avenue Suite #201     * voice:  208-522-4630

Idaho Falls, ID USA 83402-3502  * fax:    208-522-1187

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/mpich-discuss/attachments/20060307/70acd641/attachment.htm>