next up previous contents
Next: How the list of Up: Running programs in parallel Previous: How to use the

The .machines file

The .machines file generally contains one line for each machine or processor on which a process is to be started. In each line the weight (relative speed) and the name of the machine is specified in the following form:

weight:machine_name
where weight is an integer (e.g. a three times more powerful machine should have a three times higher weight).

During the run of lapw1para the file .processes is generated. This file is used by lapw2para to determine which case.vector* to read.

By default lapw1para will generate approximately 3 vector-files per processor, if enough k-points are available for distribution. The factor 3 is called ``granularity'' and should allow some load balancing in heterogeneous environments. If during siteconfig_lapw a shared memory system was selected, ``granularity'' will be set by default to 1.

For performance reasons a different ``granularity'' can be specified, by adding the line

granularity:new_granularity
to the .machines file. On shared memory architecture machines it is sometimes advisable to add a ``residue machine'' to calculate the surplus (residual) k-points (given by the expression $\mbox{MOD}(klist, \sum_j{newweight_j}$) and rely on the operating system's load balancing scheme. Such a ``residue machine'' is specified as
residue:machine_name
in the .machines file.

To distribute the remaining k-points one-by-one (and not in one junk) over all processors, the option

extrafine:1
can be set in the .machines file.


next up previous contents
Next: How the list of Up: Running programs in parallel Previous: How to use the

2000-04-11