[Q-e-developers] [Pw_forum] QE-GPU: compilation error with Espresso 5.1

Filippo Spiga spiga.filippo at gmail.com
Sun Sep 14 07:56:26 CEST 2014

Dear Mauro,

On Sep 12, 2014, at 10:30 PM, Mauro Sgroi <maurofrancesco.sgroi at gmail.com> wrote:
> I was able to compile using gfortran and the -shared flag to compile phiGEMM (I modified the file GPU/install/make_phiGEMM.inc).
> Then I was also able to linlk with openblas by manually modifying the make.sys file.
> I'm still not sure if to use the internal FFT library or if to link with a fftw3 library compiled by me.

If in the make.sys you see both -D__FFTW and -D__OMP then you are using the internal FFTW. If there is -D__FFTW3 then you are using the FFTW3 library. If --D__FFTW is specified you can avoid to put any library in the FFT_LIBS line...

> Testing the code with 8 threads (I've 8 cores, is this correct?) I got the following worrisome error:

Meaning serial (--disable-paralle) and OMP_NUM_THREADS=8? 

>  WARNING: integrated charge=   173.15368500, expected=   178.00000000
>  %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
>      Error in routine electrons (1):
>      charge is wrong
>  %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
>      stopping ...
> STOP 1
> The same inupt runs smoothly with the standard version of the code.

Serial or parallel version of the standard CPU code? Is your input require non-colin or spin magnetization = 2 ? I had user with troubles with this already in the past ...


Mr. Filippo SPIGA, M.Sc.
http://filippospiga.info ~ skype: filippo.spiga

«Nobody will drive us out of Cantor's paradise.» ~ David Hilbert

Disclaimer: "Please note this message and any attachments are CONFIDENTIAL and may be privileged or otherwise protected from disclosure. The contents are not to be disclosed to anyone other than the addressee. Unauthorized recipients are requested to preserve this confidentiality and to advise the sender immediately of any error in transmission."

More information about the developers mailing list