Hello

I have a hp station and I would like to run calculations from one model using both the two processors of my computer. Does anyone know the UNIX commands ?

Thanks

Regards

Noé

star 2

It doesn´t work. I´m using the version 3.150 of StarCD, so I have to type "Starlink" to run the calculations ("Star" works with the version 3.200). And if I type "Starlink 2", "2" becomes the name of my executable. Do I have to use the version 3.200 of StarCD, or is there an other way with the version 3.150 ? Thanks Regards Noé

for version 3.15 you have to use the prohpc command.

When the prohpc is setup, which command do I have to type on UNIX ?

You have to start the shell script casename-run.sh &

Hello, noe.. This is the steps.

From Prostar: -save model -save geometry (on buttom for STAR-HPC and put no. of processor) -save par -quit prostar

Then.. 1. open prohpc by typing prohpc (depend how u setup) 2. then flow the step in the PROHPC...

moya

Ok thanks for your help. I didn´t know I have to precise the number of processor into saving the geometry ! It´s already a good point ! Then, I don´t exactly know how to setup the prohpc, which option I have to choose. For example I don´t know how much nodes I have to enter in the "Number of nodes to be used" case ? The total number of nodes I have in my mesh ? And then, the command to run the calculations is always "starlink" ? Your help will be welcome... Regards Noé

HI,

I do not know much about this prohpc, but number of nodes will be number of processors you want to run the case on.

I bet it's well described in the manuals comming with star. Sory to say so, but RTFM.

commands to type on unix shell may be in appendix E of user guide, but I may not remember it correctly...

matej

You don't have to choose the number of processors when saving the geometry. That number is left over from early days of parallel running, but is no longer necessary. You tell it how many nodes (ie processors) to use in prohpc only.

Yeah, STAR's been way behind the pack for many years with this (PROHPC: horrible program - almost like the GUI and solver developers don't share the same office!).

But with 3.2 the decomposition's hidden at last (yippee). Let's just hope they aren't at the bottom of the same learning curve that the rest have already climbed.