[Pw_forum] Comparison of 3.1.1 and 3.2 (cvs)

Huiqun Zhou hqzhou at nju.edu.cn
Thu Nov 23 09:46:22 CET 2006


Dear list users,

10 days ago, I checked out the cvs version and compared
the speed of both versions against the same test case on
same machine.

pwscf was compiled using Intel Fortran 9.0 for x86_64, MPI
was Intel MPI-2 2.0, just the same environment for generating
the results of my previous tests on AMD opteron 285, Intel
Xeon 5060, 5140, 5150 platforms that I posted to this forum.

This test was carried on a Intel 5050 machine (dempsey 
3.0GHz/667MHz, 4GB memory). As my previous tests, the test case 
is still the one to calculate total energy of MgAl2O4 with 
calcium ferric structure, and 28 atoms in the unit cell. Here 
are the numbers:

(1) V3.1.1
1 core : 8m10s (8m09.52s)
2 cores: 4m57s (4m52.95s)
4 cores: 2m23s (2m17.68s)

(2) V3.2 (cvs, about 10 days ago)
1 core : 9m29.95s (9m11.09s)
2 cores: 6m03.24s (5m26.85s)
4 cores: 2m35.26s (2m04.52s)

There are two problems here:
(1) The differences between wall time and CPU time with version
  3.1.1 is small and seems reasonable while that between the two 
  with cvs version is relatively large. Why?
(2) Version 3.1.1 is much faster than the cvs version as the former
  took only 13 scf iterations to reach convergence while the latter
  needed 16 iterations.

Of course, it's just a cvs version and development is in progress.
I hope that developers have already noticed the difference.

Thanks to all developers for their great effort!


Huiqun Zhou








More information about the Pw_forum mailing list