CP2K-7.1 Toolchain crash at ELPA instalation

[Krzysztof Kolman]

Hello,

I have encounter a problem while trying to compile cp2k at Ubuntu 20.04 with CUDA support. When I use following toolchain command:

./install_cp2k_toolchain.sh --with-libxsmm=install --with-openblas=install --with-fftw=install --with-reflapack=no  --enable-cuda --enable-omp --gpu-ver=V100 --with-cmake=system --with-openmpi=install

I get such error:

==================== Installing ELPA ====================
elpa-2019.05.001.tar.gz is found
Installing from scratch into /cp2k-7.1/tools/toolchain/install/elpa-2019.05.001
ERROR: (./scripts/install_elpa.sh, line 103) Non-zero exit code detected.

From the make.log located at /build/elpa-2019.05.00/obj_no_thread I can see that nvcc crashes:

nvcc fatal   : Unknown option 'Wl,-rpath'

Please find attached the make. log file as well as config.log and configure.log located at /build/elpa-2019.05.00/obj_no_threa.

I think it is also worth to mention that I manage to compile cp2k with flag --with-elpa=no.

Thank you in advance for your help!

Best regards,

Krzysztof

[Tiziano Müller]

Hi Krzysztof,

please use the latest git-master of CP2K if you want to try out a
CUDA-enabled ELPA. In that version you should not only be able to build
ELPA with support for a CUDA-kernel, but CP2K should actually be able to
use it.

Best regards,
Tiziano

> --
> You received this message because you are subscribed to the Google
> Groups "cp2k" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to cp2k+uns...@googlegroups.com
> <mailto:cp2k+uns...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/cp2k/60525719-7909-44a5-b729-4314898ca143n%40googlegroups.com
> <https://groups.google.com/d/msgid/cp2k/60525719-7909-44a5-b729-4314898ca143n%40googlegroups.com?utm_medium=email&utm_source=footer>.

--
Tiziano Müller
University of Zurich
Department of Chemistry
Winterthurerstrasse 190
CH-8057 Zürich

Tel: +41 44 63 54234
www.chem.uzh.ch
tiziano...@chem.uzh.ch

[Krzysztof Kolman]

Hi Tiziano,

Thank you for your help! This time it worked :) I have one more question. After the compilation I performed the regression testing. In the beginning I got following message:

CP2K supports:  cp2kflags: omp libint fftw3 libxc acc pw_cuda xsmm dbcsr_acc spglib libvori libbqb grid_cuda

Skipping QS/regtest-cdft-hirshfeld-2 : missing required feature : parallel
Skipping QS/regtest-cdft-hirshfeld-2 : missing required feature : mpiranks>1
Skipping SIRIUS/regtest-1 : missing required feature : sirius
Skipping QS/regtest-elpa-2 : missing required feature : elpa
Skipping QMMM/QS/regtest-cdft : missing required feature : parallel
Skipping QMMM/QS/regtest-cdft : missing required feature : mpiranks==2||mpiranks==4||mpiranks==6||mpiranks==16||mpiranks==24
Skipping QS/regtest-rma : missing required feature : mpi3
Skipping QS/regtest-rma : missing required feature : mpiranks==4
Skipping QS/regtest-cdft-3 : missing required feature : parallel
Skipping QS/regtest-cdft-3 : missing required feature : mpiranks==2||mpiranks==4||mpiranks==6||mpiranks==16||mpiranks==24
Skipping QS/regtest-pexsi : missing required feature : pexsi
Skipping QS/regtest-rma-3D : missing required feature : mpi3
Skipping QS/regtest-rma-3D : missing required feature : mpiranks==8
Skipping TMC/regtest_ana_on_the_fly : missing required feature : parallel
Skipping TMC/regtest_ana_on_the_fly : missing required feature : mpiranks>2
Skipping TMC/regtest : missing required feature : parallel
Skipping TMC/regtest : missing required feature : mpiranks>1
Skipping QS/regtest-elpa-qr : missing required feature : elpa
Skipping TMC/regtest_ana_post_proc : missing required feature : parallel
Skipping TMC/regtest_ana_post_proc : missing required feature : mpiranks>1
Skipping QS/regtest-elpa : missing required feature : elpa
Skipping Fist/regtest-quip : missing required feature : quip
Skipping Fist/regtest-plumed2 : missing required feature : plumed2

Here it claims that ELPA and SIRIUS are missing. Do you know maybe why?

Best regards,

Krzysztof

[Tiziano Müller]

Hi Krzysztof,

I assume you used the `ssmp` (or the deprecated `sopt`) VERSION?
SIRIUS and ELPA are only available when built with MPI support.
So, you have to run:

make ARCH=local VERSION=psmp

Best,
Tiziano

> <https://groups.google.com/d/msgid/cp2k/60525719-7909-44a5-b729-4314898ca143n%40googlegroups.com?utm_medium=email&utm_source=footer

> <https://groups.google.com/d/msgid/cp2k/60525719-7909-44a5-b729-4314898ca143n%40googlegroups.com?utm_medium=email&utm_source=footer>>.
>
>
> --
> Tiziano Müller
> University of Zurich
> Department of Chemistry
> Winterthurerstrasse 190
> CH-8057 Zürich
>
> Tel: +41 44 63 54234

> www.chem.uzh.ch <http://www.chem.uzh.ch>
> tiziano...@chem.uzh.ch

>
> --
> You received this message because you are subscribed to the Google
> Groups "cp2k" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to cp2k+uns...@googlegroups.com
> <mailto:cp2k+uns...@googlegroups.com>.
> To view this discussion on the web visit

> https://groups.google.com/d/msgid/cp2k/c8f0d53a-bef6-44ae-ba86-e8752d0bc79dn%40googlegroups.com
> <https://groups.google.com/d/msgid/cp2k/c8f0d53a-bef6-44ae-ba86-e8752d0bc79dn%40googlegroups.com?utm_medium=email&utm_source=footer>.

[Krzysztof Kolman]

Hello Tiziano,

Thank you again for your help! After compiling psmp version of cp2k (openMPI based ), both of them, SIRIUS and ELPA are included. After the compilation I have benchmarked both of them and I got some results that I do not understand (ssmp and psmp were compiled with CUDA enabled). For testing, I was using my desktop pc, Ubuntu 20.04, Intel Core i7-4790, (4 cores, 2 threads), Gtx 1660 super and 16 gb of RAM. As benchmark I was using H2O-64.inp. Please find below what I got:

1) time cp2k.ssmp -o H2O-64.out H2O-64.inp

real    8m7,011s
user    27m34,937s
sys    1m2,519s

2) time mpirun -np 4 cp2k.psmp -o H2O-64.out H2O-64.inp

no load on GPU, only video memory used
real    49m20,818s
user    361m47,512s
sys    2m30,998s

Why do I observe such huge difference? I run it twice and I got the same result. After that I started to play little bit.

3) time mpirun -np 4 -x OMP_NUM_THREADS=2 cp2k.psmp -o H2O-64.out H2O-64.inp

real    7m59,329s
user    41m36,361s
sys    0m47,529s

4) time mpirun -np 4 --map-by slot:PE=1 -x OMP_PLACES=threads -x OMP_PROC_BIND=SPREAD -x OMP_NUM_THREADS=2 cp2k.psmp -o H2O-64.out H2O-64.inp

real    7m50,873s
user    41m25,381s
sys    0m46,038s

Tiziano, do you know why such huge discrepancies are observed? Thank you in advance for you comments!

Best regards,

Krzysztof