[Pw_forum] PW taskgroups and a large run on a BG/P

[Paolo Giannozzi]

On Feb 12, 2009, at 9:20 , Lorenzo Paulatto wrote:

> That subroutine is usually where error originated in the most  
> remotes parts of the code finally emerge. There are dozens of  
> possible errors that will not cause any apparent problem for a  
> while, until the errors accumulate enough to break positive  
> definiteness of the matrix.

this is a different case: if the subspace diagonalization it works on  
one processor,
it has to work on more that one processor as well

Paolo
---
Paolo Giannozzi, Democritos and University of Udine, Italy