[Cado-nfs-discuss] how to configure mach_desc
pierrick.gaudry at loria.fr
Sun Jun 23 09:37:29 CEST 2013
Assuming the name of the two nodes you want to use are
then you can have the following in your mach_desc:
tmpdir=/tmp <- to be adjusted !!!
bindir=/path/to/cado/build <- to be adjusted !!!
The "section" notion in the mach_desc file is just a way to group
machines that have the same set of parameters. Nothing more clever. In
particular, the choice of the name of the section is just for the human
who reads the file.
Please remember however that if you want the cadofactor.pl script to use
these machines, you should have an ssh access to them (and configured
correctly so that you don't need to type passwords).
Also, Alex Kruppa is currently rewriting the main script. The format for
the file equivalent to mach_desc will slightly change in the near future.
On Sat, Jun 22, 2013 at 09:02:56PM +0800, meng wrote:
> Hello all,
> I am puzzled abou the configure of mach_desc. Each section begins with [ clustername ]. If only a machine is used, the clustername is [local] as given by one mail.
> Now suppose I can use two computation nodes in one High computation center. Each of the nodes has two cpus. Each of the cpus has 6 cores. Each of the two nodes has 128G ram.
> How do I configure my mach_desc on the two nodes?
> I have read the sample mach_desc and some maillists. But I still dont know how to configure the mach_desc in my case.
> Thank you.
> Cado-nfs-discuss mailing list
> Cado-nfs-discuss at lists.gforge.inria.fr
More information about the Cado-nfs-discuss