Dear All, I have noticed that now the charge density files are saved by default in CP calculations and this is not affected by the disk_io input parameter. Generally this is not a problem, but it can be in certain circumstances, like restricted disk space/quota.
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. My calculation is spin-polarized and thus I have 3 such density files, and in addition I want to use the string method, which will multiply the number of these files by N-2, where N is the number of images. The mystery is further enhanced by the fact that (a) the wavefinction files (~23 MB) are dumped rather quickly (less than 30 sec) and (b) if I repeat the calculation on our pc-cluster, writing everything takes ~37 sec which is reasonable. I would appreciate your comments on: (a) What is the reasoning behind making the cp code dump density files by default? (b) Did anyone experience such anomalous I/O rates on Lemieux or any other cluster. Thanks, Silviu.
