Hi, I'm performing a scf calculation on two different architectures: one node with 60 CPU, and three node with 20 CPU every node. I noticed that on the first architecture the job run faster. I use wf_collect= .true. and outdir in my home. There is some setting to improve the calculation velocity, when I scale the job on diffent parallel nodes?
Regards Al
_______________________________________________ Pw_forum mailing list [email protected] http://pwscf.org/mailman/listinfo/pw_forum
