<html><head><meta http-equiv="content-type" content="text/html; charset=utf-8"></head><body dir="auto">I did build also with precise but did not help. The values are very wrong , not slightly. Ron<br><br><div id="AppleMailSignature" dir="ltr">Sent from my iPhone</div><div dir="ltr"><br>On Jan 7, 2019, at 20:13, Anton Kudelin <<a href="mailto:archm...@gmail.com">archm...@gmail.com</a>> wrote:<br><br></div><blockquote type="cite"><div dir="ltr"><div dir="ltr">Could you add "-fp-model precise" to CFLAGS and FCFLAGS? It won't fix 'RUNTIME FAIL', but could help with 'WRONG RESULT'.<br><br>On Monday, January 7, 2019 at 9:06:28 PM UTC+3, Ronald Cohen wrote:<blockquote class="gmail_quote" style="margin: 0;margin-left: 0.8ex;border-left: 1px #ccc solid;padding-left: 1ex;"><div style="word-wrap:break-word;line-break:after-white-space">So I tried:<div><div><br></div><div>export KMP_STACKSIZE=512M</div><div>rcohen@tomcat3:~/CP2K/cp2k$ ./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 |& tee testwith512MKMP_STACKSIZE.out &</div><div>and I still get:</div><div><br></div><div><div>< /home/rcohen/CP2K/cp2k/TEST-<wbr>Linux-x86-64-intel-psmp-2019-<wbr>01-07_18-24-16/tests/QS/<wbr>regtest-dft-vdw-corr-3 (1 of 3) done in 775.00 sec</div><div>>>> /home/rcohen/CP2K/cp2k/TEST-<wbr>Linux-x86-64-intel-psmp-2019-<wbr>01-07_18-24-16/tests/QS/<wbr>regtest-dft-vdw-corr-3</div><div>    argon05.inp                                               -85.02462435591488  WRONG RESULT TEST 1 </div><div>    argon06.inp                                               -85.18989253445228  WRONG RESULT TEST 1 </div><div>    argon07.inp                                               -85.05087192159809         RUNTIME FAIL </div><div>    argon08.inp                                               -85.05201740647929         RUNTIME FAIL </div><div>    argon09.inp                                               -85.05086520280044         RUNTIME FAIL </div><div>    argon10.inp                                               -85.05070440200512         RUNTIME FAIL </div><div>    argon11.inp                                               -84.69892988333885         RUNTIME FAIL </div><div>    argon12.inp                                               -84.69900817368848         RUNTIME FAIL </div><div>    argon13.inp                                               -84.81306482759408  WRONG RESULT TEST 1 </div><div>    argon14.inp                                               -84.69889654472566  WRONG RESULT TEST 1 </div><div>    argon-beef.inp                                            -42.46311172518392  WRONG RESULT TEST 1 </div><div>    dftd3bj_t1.inp                                             -0.00355123783846     OK (   1.19 sec) </div><div>    dftd3bj_t2.inp                                             -0.05897356220363     OK (   2.20 sec) </div><div>    dftd3bj_t3.inp                                             -0.00112424003807     OK (   3.75 sec) </div><div>    dftd3bj_t4.inp                                                -84.2983390350     OK (   3.86 sec) </div><div><<< /home/rcohen/CP2K/cp2k/TEST-<wbr>Linux-x86-64-intel-psmp-2019-<wbr>01-07_18-24-16/tests/QS/<wbr>regtest-dft-vdw-corr-3 (1 of 3) done in 775.00 sec</div><div>Starting regression tests in /home/rcohen/CP2K/cp2k/TEST-<wbr>Linux-x86-64-intel-psmp-2019-<wbr>01-07_18-24-16/tests/QS/<wbr>regtest-dft-vdw-corr-2 (2 of 3)</div><div>Starting regression tests in /home/rcohen/CP2K/cp2k/TEST-<wbr>Linux-x86-64-intel-psmp-2019-<wbr>01-07_18-24-16/tests/QS/<wbr>regtest-dft-vdw-corr-2 (2 of 3)</div><div><br></div></div><div><br></div><div>Almost all of the non vdw routines pass.</div><div><br></div><div>Sincerely,</div><div><br></div><div>Ron</div><div><br></div><div>
<div style="color:rgb(0,0,0);font-family:Helvetica;font-style:normal;font-weight:normal;letter-spacing:normal;line-height:normal;text-align:-webkit-auto;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;word-wrap:break-word;line-break:after-white-space">---<br>Ron Cohen<br><a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">rec...@gmail.com</a><br>skypename: ronaldcohen</div><div style="color:rgb(0,0,0);font-family:Helvetica;font-style:normal;font-weight:normal;letter-spacing:normal;line-height:normal;text-align:-webkit-auto;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;word-wrap:break-word;line-break:after-white-space">twitter: @recohen3<br><div><br></div></div><br><br>
</div>
<div><br><blockquote type="cite"><div>On Jan 7, 2019, at 6:12 PM, Robert Schade <<a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">robert...@uni-paderborn.<wbr>de</a>> wrote:</div><br><div><div><div><div>Signed PGP part</div><div>Could you try setting KMP_STACKSIZE to something large in the terminal<br>session with "export KMP_STACKSIZE=512m" before you rerun the regtests<br>with your intel-psmp-binary that failed before?<br>Please also make sure that the general stack size is not the problem<br>by running "ulimt -s unlimited" in the same terminal where you want to<br>execute the regtests.<br>Best Wishes<br>Robert<br><br>On 07.01.19 18:00, Ronald Cohen wrote:<br>> BTW, in case it was not clear. My Intel builds of POPT and PSMP<br>> versions were error free. The problems were all run time.<br>><br>> Ron<br>><br>> --- Ron Cohen <a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">rec...@gmail.com</a> <<a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">mailto:...@gmail.com</a>><br>> skypename: ronaldcohen twitter: @recohen3<br>><br>><br>><br>><br>>> On Jan 7, 2019, at 5:39 PM, Robert Schade<br>>> <<a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">robert...@uni-paderborn.<wbr>de</a><br>>> <<a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">mailto:rob...@uni-<wbr>paderborn.de</a>>> wrote:<br>>><br>>> Signed PGP part r is automatically private because it is the<br>>> first iteration variable. Every drho(s, i) is only read and<br>>> written in exactly one loop iteration. The statement<br>>> "COLLAPSE(3)" collapses the three perfectly nested loops into one<br>>> loop. So, IMHO, this code looks ok. Best Wishes Robert<br>>><br>>><br>>> On 07.01.19 14:52, Ronald Cohen wrote:<br>>>> Yes, I agree. I have tried the 2018.05 and the 2019.1 intel<br>>>> compilers. The POPT version runs fine, but the PSMP version<br>>>> fails in the vDW routines. I find things like: in<br>>>> qs_dispersion_nonloc.F<br>>>><br>>>> !$OMP PARALLEL DO DEFAULT(NONE)<span>                 <wbr>    </span> & !$OMP<br>>>> SHARED(ispin,i,n,lo,drho,drho_<wbr>r)<span>  </span> & !$OMP<br>>>> PRIVATE(s) & !$OMP<span>            </span> COLLAPSE(3) DO r = 0, n(3)-1 DO<br>>>> q = 0, n(2)-1 DO p = 0, n(1)-1 s = r*n(2)*n(1)+q*n(1)+p+1<br>>>> drho(s, i) = drho(s, i)+drho_r(i, ispin)%pw%cr3d(p+lo(1), q<br>>>> +lo(2), r+lo(3)) END DO END DO END DO !$OMP END PARALLEL DO END<br>>>> DO END DO<br>>>><br>>>> Doesn’t this have to be marked as a reduction? And shouldn’t r,<br>>>> q, p be labeled private? Perhaps this is automatic, but I do<br>>>> not see that said anywhere. Does gnu treat such differently<br>>>> than intel? Just ideas.<br>>>><br>>>> I am currently trying the toolchain, but it is building<br>>>> everything from scratch, including blas, lapack, scalapack etc<br>>>> etc, so will take days.<br>>>><br>>>> Thank you for your help,<br>>>><br>>>> Sincerely,<br>>>><br>>>> Ron<br>>>><br>>>> --- Ron Cohen <a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">rec...@gmail.com</a> <<a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">mailto:...@gmail.com</a>><br>>> <<a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">mailto:...@gmail.com</a>><br>>>> skypename: ronaldcohen twitter: @recohen3<br>>>><br>>>><br>>>><br>>>><br>>>>> On Jan 7, 2019, at 2:16 PM, Robert Schade<br>>>>> <<a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">robert...@uni-paderborn.<wbr>de</a><br>>>>> <<a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">mailto:rob...@uni-<wbr>paderborn.de</a>><br>>>>> <<a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">mailto:rob...@uni-<wbr>paderborn.de</a>>> wrote:<br>>>>><br>>>>> Building cp2k on Intel Xeon Phi Knights Landing (KNL, not to<br>>>>> be confused with KNC!) is not different from building it on<br>>>>> any other Intel CPU. Hence, I think that the failing regtests<br>>>>> point to an underlying issue. Which exact version of the<br>>>>> Intel Compiler and MKL have you tried? Best Wishes Robert<br>>>>><br>>>>> On 06.01.19 01:59, Ronald Cohen wrote:<br>>>>>> OK—sorry for all the noise. I am trying:<br>>>>>> ./install_cp2k_toolchain.sh --with-elpa=install<br>>>>>> --with-libint=install --with-gcc=install I hate not being<br>>>>>> able to use my intel tools which work for me for everything<br>>>>>> else just fine.<br>>>>>><br>>>>>> Ron<br>>>>>><br>>>>><br>>>>> -- Robert Schade Paderborn Center for Parallel Computing<br>>>>> (PC2) University of Paderborn Warburger Str. 100 D-33098<br>>>>> Paderborn Germany <a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">robe...@uni-paderborn.de</a><br>>> <<a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">mailto:rob...@uni-<wbr>paderborn.de</a>><br>>>>> <<a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">mailto:rob...@uni-<wbr>paderborn.de</a>> +49/(0)5251/60-5393<br>>>>><br>>>>> -- You received this message because you are subscribed to a<br>>>>> topic in the Google Groups "cp2k" group. To unsubscribe from<br>>>>> this topic, visit<br>>>>> <a href="https://groups.google.com/d/topic/cp2k/gzmRqKNt62U/unsubscribe" target="_blank" rel="nofollow">https://groups.google.com/d/<wbr>topic/cp2k/gzmRqKNt62U/<wbr>unsubscribe</a>.<br>>><br>>>>><br>>> To unsubscribe from this group and all its topics, send an email<br>>>>> to <a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">cp2k+uns...@googlegroups.<wbr>com</a><br>>> <<a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">mailto:cp2...@<wbr>googlegroups.com</a>>. To post to this<br>>> group, send<br>>>>> email to <a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">cp...@googlegroups.com</a><br>>>>> <<a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">mail...@googlegroups.com</a>><wbr>.<br>>> Visit this group at<br>>>>> <a href="https://groups.google.com/group/cp2k" target="_blank" rel="nofollow">https://groups.google.com/<wbr>group/cp2k</a>. For more options,<br>>>>> visit <a href="https://groups.google.com/d/optout" target="_blank" rel="nofollow">https://groups.google.com/d/<wbr>optout</a>.<br>>>><br>>>> -- You received this message because you are subscribed to the<br>>>> Google Groups "cp2k" group. To unsubscribe from this group and<br>>>> stop receiving emails from it, send an email to<br>>>> <a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">cp2k+uns...@googlegroups.<wbr>com</a><br>>> <<a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">mailto:cp2...@<wbr>googlegroups.com</a>><br>>>> <<a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">mailto:cp2...@<wbr>googlegroups.com</a>>. To post to this<br>>>> group, send email to <a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">cp...@googlegroups.com</a><br>>>> <<a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">mail...@googlegroups.com</a>> <<a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">mail...@googlegroups.com</a>><wbr>.<br>>>> Visit this group at <a href="https://groups.google.com/group/cp2k" target="_blank" rel="nofollow">https://groups.google.com/<wbr>group/cp2k</a>. For<br>>>> more options, visit <a href="https://groups.google.com/d/optout" target="_blank" rel="nofollow">https://groups.google.com/d/<wbr>optout</a>.<br>>><br>>> -- Robert Schade Paderborn Center for Parallel Computing (PC2)<br>>> University of Paderborn Warburger Str. 100 D-33098 Paderborn<br>>> Germany <a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">robe...@uni-paderborn.de</a><br>>> <<a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">mailto:rob...@uni-<wbr>paderborn.de</a>> +49/(0)5251/60-5393<br>>><br>><br>> -- You received this message because you are subscribed to the<br>> Google Groups "cp2k" group. To unsubscribe from this group and stop<br>> receiving emails from it, send an email to<br>> <a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">cp2k+uns...@googlegroups.<wbr>com</a><br>> <<a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">mailto:cp2...@<wbr>googlegroups.com</a>>. To post to this group,<br>> send email to <a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">cp...@googlegroups.com</a><br>> <<a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">mail...@googlegroups.com</a>><wbr>. Visit this group at<br>> <a href="https://groups.google.com/group/cp2k" target="_blank" rel="nofollow">https://groups.google.com/<wbr>group/cp2k</a>. For more options, visit<br>> <a href="https://groups.google.com/d/optout" target="_blank" rel="nofollow">https://groups.google.com/d/<wbr>optout</a>.<br><br>--<br>Robert Schade<br>Paderborn Center for Parallel Computing (PC2)<br>University of Paderborn<br>Warburger Str. 100<br>D-33098 Paderborn<br>Germany<br><a target="_blank" gdf-obfuscated-mailto="amaOaVKiDQAJ" rel="nofollow">robe...@uni-paderborn.de</a><br>+49/(0)5251/60-5393</div></div><br></div></div></blockquote></div><br></div></div></blockquote></div>

<p></p>

-- <br>
You received this message because you are subscribed to a topic in the Google Groups "cp2k" group.<br>
To unsubscribe from this topic, visit <a href="https://groups.google.com/d/topic/cp2k/gzmRqKNt62U/unsubscribe">https://groups.google.com/d/topic/cp2k/gzmRqKNt62U/unsubscribe</a>.<br>
To unsubscribe from this group and all its topics, send an email to <a href="mailto:cp2k+uns...@googlegroups.com">cp2k+uns...@googlegroups.com</a>.<br>
To post to this group, send email to <a href="mailto:cp...@googlegroups.com">cp...@googlegroups.com</a>.<br>
Visit this group at <a href="https://groups.google.com/group/cp2k">https://groups.google.com/group/cp2k</a>.<br>
For more options, visit <a href="https://groups.google.com/d/optout">https://groups.google.com/d/optout</a>.<br>
</div></blockquote></body></html>