NPAR problem with VASP 5.2.11 vs. 4.6.36

Message

alex · #1 Post by **alex** » Wed Mar 16, 2011 12:55 pm

Hi there,

I'm doing a frequency calculation with VASP 5.2.11. However, using this version I receive an error:

--snip--
RMM: 34 -0.857539860988E+03 -0.47479E-05 -0.78739E-07 1112 0.132E-03 0.531E-04
RMM: 35 -0.857539861532E+03 -0.54488E-06 -0.25562E-07 918 0.802E-04
1 F= -.85753986E+03 E0= -.85753986E+03 d E =-.222708E-13 mag= 0.0000
Finite differences POTIM= 0.02000 DOF= 12

-----------------------------------------------------------------------------
| |
| EEEEEEE RRRRRR RRRRRR OOOOOOO RRRRRR ### ### ### |
| E R R R R O O R R ### ### ### |
| E R R R R O O R R ### ### ### |
| EEEEE RRRRRR RRRRRR O O RRRRRR # # # |
| E R R R R O O R R |
| E R R R R O O R R ### ### ### |
| EEEEEEE R R R R OOOOOOO R R ### ### ### |
| |
| VASP internal routines have requested a change of the k-point set. |
| Unfortunately this is only possible if NPAR=number of nodes. |
| Please remove the tag NPAR from the INCAR file and restart the |
| calculations. |
| |
| ----> I REFUSE TO CONTINUE WITH THIS SICK JOB ..., BYE!!! <---- |
| |
-----------------------------------------------------------------------------

I'd have to unset NPAR loosing about 50% performance. Running the job with the latest VASP 4.6 works fine.

Space group is P1 from start, so no symmetry issues. I'm using the gamma-Version of the code, so I also wonder what the change of k-point set is about.

Thanks,

Alex

boris · #2 Post by **boris** » Wed Mar 16, 2011 9:36 pm

Sorry, this won't help you but...

I so love vasp error messages... They are always so useful...

Boris

alex · #3 Post by **alex** » Thu Mar 17, 2011 7:38 am

Yep, Georg probably loves it, too. :-)

And the NPAR issue? Sometime i _have_ to use version 5 ...

Thanks,

Alex

forsdan · #4 Post by **forsdan** » Fri Mar 18, 2011 2:14 pm

The difference between vasp 4.6 and vasp 5 is that vasp 5 changes the k-point mesh on the fly to adopt to the symmetry of the displaced geometry. The ability to do so is dependent on the memory distribution, which is why you have to set the NPAR tag to the default value in order set the distribution correctly.

My guess to why you still get a problem if you only have one k-point, is that vasp is probably calling the change k-point routine and checks the memory distribution mode before it checks that you only have one point. But I might be wrong. I haven't looked into the gamma-point only mode.

Cheers
/Dan

<span class='smallblacktext'>[ Edited Fri Mar 18 2011, 03:18PM ]</span>

alex · #5 Post by **alex** » Fri Mar 18, 2011 4:24 pm

Thanks you for the explanation, Dan! It's not a bug, it's a feature. ;-)
Perhaps somebody out of the VASP team could change this for the gamma-point version, please please please.

Cheers,

Alex