[Beowulf] RE: Compare and contrast MPI implementations

[Robert G. Brown]

On Mon, 19 Dec 2005, Leif Nixon wrote:

> "Robert G. Brown" <rgb at phy.duke.edu> writes:
>
>> On Fri, 16 Dec 2005, Leif Nixon wrote:
>>
>>> But it doesn't help for David's problem; that environment variables
>>> aren't passed across rsh.
>>
>> Making it a GREAT time to point out that environment variables CAN be
>> passed across ssh.  See man ssh, ~/.ssh/environment.
>
> Well, that facility's usefulness is rather limited, since it's global
> across all the user's jobs.
>
>

But there are so many different ways to remove fur from felines.  E.g.

rgb at lilith|B:1001>ssh lucifer "FROG='tasty';echo \$FROG"
tasty

Or shell scripts.  Or conditionals in your shell init files (or sshrc)
such as (pseudocode):

if(FROG == 'tasty')
   MPI_FLAVOR=tasty
   export MPI_FLAVOR
else
   MPI_FLAVOR=toasty
   export MPI_FLAVOR

together with the example above, where of course you can do anything you
like under the various conditionals.  Seriously.  Not all solutions are
necessarily "easy" to make automagical for a non-Unix-hacker type person
or easy to administratively support, but neither are they particularly
difficult and they CAN be documented, one would think.

But I think modules already sound like they take care of the problem.  I
was just advertising for ssh, since I like smart remote shells.  rsh is
fast, but it isn't smart and it isn't secure.

   rgb

-- 
Robert G. Brown	                       http://www.phy.duke.edu/~rgb/
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone: 1-919-660-2567  Fax: 919-660-2525     email:rgb at phy.duke.edu