You simply cannot get 2 pico second trajectory in 10 steps! Your timestep may
at most be 1fs.
Huiqun Zhou
-- Original --
From: "Narendranath Ghosh";
Date: Fri, Jul 22, 2016 02:17 AM
To: "pw_forum";
Subject:
Masoud,
While you used generic BLAS/LAPACK libraries when you compiled your
qe 4.2, you were using MKL in compiling qe 4.3.1. But the compiler
never found where the MKL libraries are. How did you set the environment
for MKL? You should "source" the mklvarsem64t.sh in .bashrc or
.bash_profile as
Hi,
You have 32 CPU cores that devided into 4 pools, so you have only
8 CPU cores for each pool. 8 is not a square of any integer, and
it's why it used serial algorithm.
you need to run
mpirun -np 32 pw.x -npool 2 -ndiag 4
BUT, you'd better make tests to determine the best combination.
For my
Paolo,
I have been using Intel MPI these years, from its 2.x, 3.1 to
3.2 now. I use it with two job schedulers, torque and lsf, on
two clusters of course. At least, there are no big problems.
I had ever tried to switch to openmpi, but found it was slower
than Intel MPI at that time (about 3
It simply means that using pools in phonon calculation is meaningless.
But it may not be the main cause of your error.
Dr. Huiqun Zhou
@Earth Sciences, Nanjing University, China
Axel,
Thank you very much for your response!
> please note that benchmark numbers are useless without
> offering a way to reproduce them. so please provide the
> the input used for this test.
>
Sorry, I'm out of town right now, I'll send the input file to you
directly as soon as I return home.
Your system should have only one atom and then you can get a slightly
smaller total and absolute magnetization than Fe-bcc.
Dr. Huiqun Zhou
@Nanjing University, China
> -Original Message-
> ???: Hongbin Zhang
> ??: Wed, 23 Jan 2008 19:19:09 +0100
> ???:
> ??: [Pw_forum] nonmagnetic