[Pw_forum] problems with GPU-accelerated Quantum ESPRESSO

[Filippo Spiga]

Dear Afanasiev Ilya,

like you many people do a simple mistake while they configure QE: adding macros to the configure command line without knowing how internally works.

Please run ./configure WITHOUT any CPPFLAGS or CCFLAGS or DFLAGS. 

Only AFTER the configure process (if the configure did not detect your environment properly) edit manually make.sys

And if for some reason the configure does not work please let us know so we can fix it.

Cheers,
F

On Mar 7, 2014, at 6:55 PM, Илья Афанасьев <afanasiev_ilya at icloud.com> wrote:

> Dear Sirs,
> 
> May I ask here some questions about GPU Quantum ESPRESSO installation?
> 
> If I may, here is my problem.
> 
> I`m installing Quantum GPU-accelerated ESPRESSO on the «Lomonosov» supercomputer in Russia, You can find us in top500 list on the 38 position, in November 2013 list.
> 
> So, in the beginning I installed the version without GPU, and I didn`t have any problems.
> 
> If it is important, I used such configure command:
> 
> ./install/configure --with-internal-lapack --with-internal-blas CC=icc FC=ifort F77=ifort CCFLAGS=-openmp CPPFLAGS=-openmp --enable-parallel --enable-openmp
> 
> After that I tried to compile GPU part, just as in the README file
> 
> I used such configure command:
> 
> ./configure --enable-parallel --enable-openmp --with-scalapack --enable-cuda --with-cuda-dir=/opt/cuda --disable-magma DFLAGS=-DMPICH_SKIP_MPICXX
> because I`m using Intel compilers and IntelMPI
> 
> But, during the compiling process, I had some errors, here they are:
> 
> /opt/cuda-5.0/bin/nvcc -O3 -gencode arch=compute_13,code=sm_13 -gencode arch=compute_20,code=sm_20 -gencode arch=compute_20,code=sm_21 -gencode arch=compute_30,code=sm_30 -gencode arch=compute_30,code=sm_35   -I../../include -I../include  -I/opt/ccoe/espresso-5.0.2/GPU/..//phiGEMM/include -I/opt/cuda-5.0/include -DMPICH_SKIP_MPICXX    -c cuda_init.cu
>                                   cuda_init.cu(361): error: identifier "ngpus_per_process" is undefined
>                                   cuda_init.cu(363): error: identifier "qe_gpu_bonded" is undefined
>                                   cuda_init.cu(376): error: identifier "qe_gpu_kernel_launch" is undefined
>                                   cuda_init.cu(376): error: identifier "update_gpu_kernel_specs" is undefined
>                                   cuda_init.cu(413): warning: variable "lRankThisNode" was declared but never referenced
>                                   cuda_init.cu(413): warning: variable "lSizeThisNode" was declared but never referenced
>                                   4 errors detected in the compilation of "/tmp/
> 
> When I looked into the cuda_init.cu file, I saw such code:
> 
> #if defined(__CUDA)
> 
> long ngpus_detected;
> long ngpus_used;
> long ngpus_per_process;
> long procs_per_gpu;
> 
> #endif
> 
> and after this using this variables without #if defined #endif
> 
> So I`m sure that __CUDA flag is not set, but why?
> 
> 								Sincerely, 
> 									Afanasiev Ilya, student of Moscow State University.
> 
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum

--
Mr. Filippo SPIGA, M.Sc.
http://www.linkedin.com/in/filippospiga ~ 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."