Paolo Giannozzi wrote: > On Wednesday 08 February 2006 12:18, Silviu Zilberman wrote: > > >> I have been running calculations on Lemieux which is an alpha cluster >> super computer in Pittsburgh. For some reason that is still mysterious >> to me, writing these density files on the scratch space took very long >> time, ~30 (!) minutes for a 68MB file. >> > > maybe you should rename your machine "Lepire" :-) > > The charge density file should be written only when the wavefunctions > are written, every "isave" steps and at the end of the run. If it is written > at each time step, this is definitely wrong. > The charge density is written only every isave steps, and I set it to a very large number to avoid this time-consuming i/o. But even if I write it just once at the end of the calculation, it would still require ~90 minutes for 3 files, which is completely crazy, given that the maximal time allocated per job is 12 hours on this supercomputer. > The charge density in a parallel calculation is collected to a single node > and written from there. Since it is not wise to collect it into a single > array, each slice from each processor is collected and written. Maybe > this algorithm is not optimal (maybe it is even "pessimal"). You should > try to understand where exactly the machine spends all this time and > why > I may do it, but for the time being, these files are not very useful for me. I can change the code to respect again the disk_io parameter and avoid writing these files all together. However I would like to know first if there was some reasoning behind dumping these files by default without user control over it.
Thanks, Silviu.
