[CP2K-user] van der Waals regtests fail on Intel KNL, and build glitches
Anton Kudelin
archm... at gmail.com
Fri Jan 4 06:33:36 UTC 2019
Dear Ronald,
Could you attach your arch file, please? Did you try to build CP2K via the
toolchain?
Anton K.
On Thursday, January 3, 2019 at 11:59:33 PM UTC+3, Ronald Cohen wrote:
>
> regtest-dft-vdw-corr-2 and regtest-dft-vdw-corr-3 regtests fail badly with
> PSMP under the intelx arch file or intel.psmp .
>
> starting regression tests in
> /home/rcohen/CP2K/cp2k/TEST-Linux-x86-64-intel-psmp-2019-01-03_19-57-15/tests/QS/regtest-dft-vdw-corr-3
> (1 of 3)
> >>>
> /home/rcohen/CP2K/cp2k/TEST-Linux-x86-64-intel-psmp-2019-01-03_19-57-15/tests/QS/regtest-dft-vdw-corr-3
> argon05.inp
> -85.02651698126142 WRONG RESULT TEST 1
> argon06.inp
> -85.18843047970840 WRONG RESULT TEST 1
> argon07.inp
> -85.05106863151636 RUNTIME FAIL
> argon08.inp
> -85.05196996972870 RUNTIME FAIL
> argon09.inp
> -85.05093416391124 RUNTIME FAIL
> argon10.inp
> -85.05042273661144 RUNTIME FAIL
> argon11.inp
> -84.69895559036574 RUNTIME FAIL
> argon12.inp
> -84.69905608268989 RUNTIME FAIL
> argon13.inp
> -84.81586971061908 WRONG RESULT TEST 1
> argon14.inp
> -84.69906586167036 WRONG RESULT TEST 1
> argon-beef.inp
> -42.46319102018110 WRONG RESULT TEST 1
> dftd3bj_t1.inp
> -0.00355123783846 OK ( 2.04 sec)
> dftd3bj_t2.inp
> -0.05897356220363 OK ( 3.56 sec)
> dftd3bj_t3.inp
> -0.00112424003807 OK ( 4.34 sec)
> dftd3bj_t4.inp
> -84.2983390350 OK ( 4.39 sec)
> <<<
> /home/rcohen/CP2K/cp2k/TEST-Linux-x86-64-intel-psmp-2019-01-03_19-57-15/tests/QS/regtest-dft-vdw-corr-3
> (1 of 3) done in 1253.00 sec
> Starting regression tests in
> /home/rcohen/CP2K/cp2k/TEST-Linux-x86-64-intel-psmp-2019-01-03_19-57-15/tests/QS/regtest-dft-vdw-corr-2
> (2 of 3)
> >>>
> /home/rcohen/CP2K/cp2k/TEST-Linux-x86-64-intel-psmp-2019-01-03_19-57-15/tests/QS/regtest-dft-vdw-corr-2
> dftd3_t4.inp
> -0.02290735609940 OK ( 2.99 sec)
> dftd3_t5.inp
> -0.00012095226817 OK ( 2.60 sec)
> dftd3_t6.inp
> -84.2986351038 OK ( 4.46 sec)
> dftd3_t7.inp
> -84.2986244998 OK ( 4.21 sec)
> dftd3_t8.inp
> -0.00115782786021 OK ( 4.42 sec)
> dftd3_t9.inp
> -0.00115772641844 OK ( 2.57 sec)
> dftd3_t10.inp
> -0.00117296325816 OK ( 2.65 sec)
> dftd3_t11.inp
> -84.2986400085 OK ( 14.46 sec)
> dftd3_t12.inp
> -0.00117296325816 OK ( 6.65 sec)
> dftd3_t13.inp
> -0.00045037056834 OK ( 1.77 sec)
> dftd3_t14.inp
> -0.00022349123009 OK ( 1.83 sec)
> argon01.inp
> -85.04150374301580 WRONG RESULT TEST 1
> argon02.inp
> -85.20388952727301 WRONG RESULT TEST 1
> argon03.inp
> -84.83719734168366 WRONG RESULT TEST 1
> argon04.inp
> -84.50083943001719 WRONG RESULT TEST 1
>
> This being run with:
>
> ./tools/regtesting/do_regtest -arch Linux-x86-64-intel -version psmp
> -restrictdir QS/regtest-dft-vdw-corr-1/ -restrictdir
> QS/regtest-dft-vdw-corr-2/ -restrictdir QS/regtest-dft-vdw-corr-3/
> -restrictdir QS/regtest-dft-vdw-corr-3/ -nobuild -mpiranks 4 -ompthreads 4
> -maxtasks 16
>
> and was compiled with
> make -j 8 ARCH=Linux-x86-64-intelx VERSION=psmp SYM=1
> LIBINTROOT=/mnt/beegfs LIBXCROOT=/mnt/beegfs ELPAROOT=/mnt/beegfs AVX=3
> MIC=0 CC=icc CXX=icpc OMP=1 IPO=0
>
> There are also a few glitches in the build. In particular I had to
> cp ./lib/Linux-x86-64-intelx/psmp/libxsmm/include/*
> ./obj/Linux-x86-64-intelx/psmp/
> and
> similarly for libxsmm mod files as the build process does not include them
> automatically even after building then via the arch file. Also
> -j n does not work -- one gets:
> make[3]: warning: jobserver unavailable: using -j1. Add '+' to parent
> make rule
>
> Any help appreciated!
>
> Sincerely,
>
> Ronald Cohen
>
> Extreme Materials Initiative
> Geophysical Laboratory
> Carnegie Institution
> 5251 Broad Branch Rd., N.W.
> Washington, D.C. 20015
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20190103/72b9f46c/attachment.htm>
More information about the CP2K-user
mailing list