[Wien] k-parallel execution II

[Bertoni Giovanni]

First: Sorry gelfand, I have done an auto-reply and I sent you directly
the mail istead to the wien-list!

second: Dear users, 

putting in the .machine:

1:localhost
1:localhost

for a 2 processors machine, I have the error

STDOUT 
----------------------------------
LAPW0 END
localhost: Connection refused
localhost: Connection refused
-----------------------------------

is it a probem of configuration?
please tell me.
thank you
giovanni

gelfand a écrit :
> 
> This is very easy to do.
> In the .machines file, just put
> 1:localhost
> 1:localhost
> 1:localhost
> <<as many lines as you have processors>>
> This only makes sense if you are using single-threaded BLAS/LAPACK
> libraries.  For dual-CPU machines it is definitely the way to go.
> For a 16-way machine, I'm not sure, it depends on how much RAM
> is available per CPU (you certainly don't want to cause the machine
> to swap to disk).
> 
> >===== Original Message From Bertoni Giovanni <giovanni at cemes.fr> =====
> >Dear users,
> >
> >I have a question about k-parallelization:
> >
> >is it possible to do a k-parallelization on a single machine with
> >multiple processors?
> >for exemple, I try on a machine with >= 8 processors.
> >I have tried with the TiC example (47 k-points) in interactive mode (no
> >scripts or batch files...)
> >
> >this is the .machine file
> >-----------------------------------------------
> ># .machines is the control file for parallel execution. Add lines like
> >#
> >#   speed:machine_name
> >#
> ># for each machine specifying there relative speed. For mpi
> >parallelization use
> >#
> >#   speed:machine_name:1 machine_name:1
> >#   lapw0:machine_name:1 machine_name:1
> >#
> ># further options are:
> >#
> >#   granularity:number (for loadbalancing on irregularly used machines)
> >#   residue:machine_name  (on shared memory machines)
> >#   extrafine         (to distribute the remaining k-points one after
> >the other)
> >#
> ># granularity sets the number of files that will be approximately
> ># be generated by each processor; this is used for load-balancing.
> ># On very homogeneous systems set number to 1
> ># if after distributing the k-points to the various machines residual
> ># k-points are left, they will be distributed to the
> >residual-machine_name.
> >#
> >1:localhost:8
> >lapw0:localhost:8
> >--------------------------------------------------
> >
> >..very simple. and this is the TiC.dayfile
> >
> >--------------------------------------------------
> >...
> >    start      (Tue Nov 25 18:04:11 MET 2003) with lapw0 (20/20 to go)
> >>   lapw0 -p   (18:04:11) starting parallel lapw0 at Tue Nov 25 18:04:12 MET
> 2003
> >-------- .machine1 : 8 processors
> >localhost:8
> >--------
> >**  lapw0 crashed!
> >0.0u 0.3s 0:01 24% 0+0k 10+3io 0pf+0w
> >
> >>   stop error
> >---------------------------------------------------
> >
> >
> >if i don't use parallel lapw0, I have the crash in lapw1. the test was
> >done on a machine that has the mpirun.
> >please tell me.
> >if is not possible to do a k-parallelization on a single machine, I will
> >use a fine grained parallelization, and I will have another question for
> >you... (next time)
> >Thank you.
> >giovanni.
> >_______________________________________________
> >Wien mailing list
> >Wien at zeus.theochem.tuwien.ac.at
> >http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> 
> Martin Gelfand
> 
> Associate Professor.............Phone: 970 491 5263
> Department of Physics.............Fax: 970 491 7947
> Colorado State University.......Email: gelfand at lamar.colostate.edu
> Fort Collins CO 80523-1875