Dear Psi4 developers
I have recently started using Psi4, and am impressed with the feature set, and ease of use. I have however, noticed that solvation calculations (through the external PCMSolver) don’t seem to be OPENMP parallelized, only running on a single core. Are there any plans to parallelize this in the future?
Kind regards
Chris
Hmm, calling PCM shouldn’t cause Psi4 to run on only one core. Is this just that PCMSolver does not appear to be parallelized?
There should be a timer.dat built when an infield is run. Would you mind posting that here.
I agree with the assessment of @dgasmith What PCMSolver does, once per SCF iteration, is to update the apparent surface charge and that is one matrix-vector multiplication per iteration. That is not parallelized, but it should only be a fraction of the cost of the SCF calculation itself. I am curious to see the timer.dat file.
I have performed two SCF energy calculations on the same molecule; one with PCM, and the other with no PCM. Oddly, the former does not generate a timer.dat file (the latter does). Never-the-less, I have included the output for the PCM calculation. The PCM calculation took 1271 seconds total time vs 43 seconds for the gas-phase calculation.
Kind regards
Chris
PCM Calculation:
----------------------------------------------------------------------- Psi4: An Open-Source Ab Initio Electronic Structure Package Psi4 1.1 release Git: Rev {} add49b9 R. M. Parrish, L. A. Burns, D. G. A. Smith, A. C. Simmonett, A. E. DePrince III, E. G. Hohenstein, U. Bozkaya, A. Yu. Sokolov, R. Di Remigio, R. M. Richard, J. F. Gonthier, A. M. James, H. R. McAlexander, A. Kumar, M. Saitow, X. Wang, B. P. Pritchard, P. Verma, H. F. Schaefer III, K. Patkowski, R. A. King, E. F. Valeev, F. A. Evangelista, J. M. Turney, T. D. Crawford, and C. D. Sherrill, J. Chem. Theory Comput. 13(7) pp 3185--3197 (2017). (doi: 10.1021/acs.jctc.7b00174) ----------------------------------------------------------------------- Psi4 started on: Monday, 29 January 2018 08:19AM Process ID: 19919 PSIDATADIR: /share/apps/bin/psi4/share/psi4 Memory: 500.0 MiB Threads: 16 ==> Input File <== -------------------------------------------------------------------------- memory 20480 mb molecule Rhcomplex { -1 1 Rh 0.006889057 -0.012070350 -0.027437431 O -0.021667629 1.588200281 1.176423257 O 0.988455214 -1.073984850 1.350183427 C -0.066088812 1.422843234 2.578468904 C 0.707675562 -1.005035450 2.628624214 C -0.388437545 -0.012066213 3.016927906 O 1.291306990 -1.664104499 3.471345349 H -1.348867844 -0.321140585 2.575539875 H -0.487777557 -0.050229232 4.110210695 O -1.061481812 0.983468856 -1.407085364 O 0.001598449 -1.651338515 -1.220080046 C -0.896230180 0.802181568 -2.695011301 C 0.139857385 -1.527929476 -2.620489679 C 0.295257815 -0.073606752 -3.090055697 O -1.641574041 1.269646565 -3.537794502 H 1.219693537 0.350229825 -2.670316593 H 0.376335275 -0.049058429 -4.185591274 O -1.838185033 -0.841366791 0.318619314 O 1.863006355 0.743407069 -0.476102785 H -1.619623230 -1.554841659 -0.336608880 H -2.388762071 -0.190994351 -0.164249189 H 1.969885034 1.334339120 0.294380307 H 2.463155006 -0.032936058 -0.387827587 H -0.764241538 -1.943629046 -3.110354055 H 0.922486530 1.683022901 3.014940882 H 1.037798977 -2.063898807 -2.890374420 H -0.855819050 2.071556609 2.927606569 } pcm = { Units = Angstrom Medium { SolverType = IEFPCM Solvent = Water } Cavity { RadiiSet = UFF Type = GePol Scaling = True Area = 0.3 Mode = Implicit } } set globals { guess SAD OPT_COORDINATES both basis x2c-svpall basis_relativistic x2c-svpall df_basis_scf x2c-coulomb-fit relativistic x2c MAXITER 256 dft_basis_tolerance 1.0E-11 pcm true } energy('B3LYP-D3MBJ') #optimize('B3LYP-D3MBJ') -------------------------------------------------------------------------- Memory set to 19.073 GiB by Python driver. *** tstart() called on compute-0-27.local *** at Mon Jan 29 08:19:38 2018 => Loading Basis Set <= Name: X2C-SVPALL Role: ORBITAL Keyword: BASIS atoms 1 entry RH line 1583 file /share/apps/cperry/bin/psi4/share/psi4/basis/x2c-svpall.gbs atoms 2-3, 7, 10-11, 15, 18-19 entry O line 121 file /share/apps/cperry/bin/psi4/share/psi4/basis/x2c-svpall.gbs atoms 4-6, 12-14 entry C line 81 file /share/apps/cperry/bin/psi4/share/psi4/basis/x2c-svpall.gbs atoms 8-9, 16-17, 20-27 entry H line 16 file /share/apps/cperry/bin/psi4/share/psi4/basis/x2c-svpall.gbs --------------------------------------------------------- SCF by Justin Turney, Rob Parrish, Andy Simmonett and Daniel Smith RKS Reference 16 Threads, 19531 MiB Core --------------------------------------------------------- ==> Geometry <== Molecular point group: c1 Full point group: C1 Geometry (in Angstrom), charge = -1, multiplicity = 1: Center X Y Z Mass ------------ ----------------- ----------------- ----------------- ----------------- RH 0.032139469495 0.042136477815 0.007555036500 102.905504292000 O 0.003582783495 1.642407108815 1.211415724500 15.994914619560 O 1.013705626495 -1.019778022185 1.385175894500 15.994914619560 C -0.040838399505 1.477050061815 2.613461371500 12.000000000000 C 0.732925974495 -0.950828622185 2.663616681500 12.000000000000 C -0.363187132505 0.042140614815 3.051920373500 12.000000000000 O 1.316557402495 -1.609897671185 3.506337816500 15.994914619560 H -1.323617431505 -0.266933757185 2.610532342500 1.007825032070 H -0.462527144505 0.003977595815 4.145203162500 1.007825032070 O -1.036231399505 1.037675683815 -1.372092896500 15.994914619560 O 0.026848861495 -1.597131687185 -1.185087578500 15.994914619560 C -0.870979767505 0.856388395815 -2.660018833500 12.000000000000 C 0.165107797495 -1.473722648185 -2.585497211500 12.000000000000 C 0.320508227495 -0.019399924185 -3.055063229500 12.000000000000 O -1.616323628505 1.323853392815 -3.502802034500 15.994914619560 H 1.244943949495 0.404436652815 -2.635324125500 1.007825032070 H 0.401585687495 0.005148398815 -4.150598806500 1.007825032070 O -1.812934620505 -0.787159963185 0.353611781500 15.994914619560 O 1.888256767495 0.797613896815 -0.441110317500 15.994914619560 H -1.594372817505 -1.500634831185 -0.301616412500 1.007825032070 H -2.363511658505 -0.136787523185 -0.129256721500 1.007825032070 H 1.995135446495 1.388545947815 0.329372774500 1.007825032070 H 2.488405418495 0.021270769815 -0.352835119500 1.007825032070 H -0.738991125505 -1.889422218185 -3.075361587500 1.007825032070 H 0.947736942495 1.737229728815 3.049933349500 1.007825032070 H 1.063049389495 -2.009691979185 -2.855381952500 1.007825032070 H -0.830568637505 2.125763436815 2.962599036500 1.007825032070 Running in c1 symmetry. Rotational constants: A = 0.03362 B = 0.01212 C = 0.01116 [cm^-1] Rotational constants: A = 1007.84238 B = 363.28486 C = 334.45463 [MHz] Nuclear repulsion = 2018.598285505243894 Charge = -1 Multiplicity = 1 Electrons = 158 Nalpha = 79 Nbeta = 79 ==> Algorithm <== SCF Algorithm Type is DF. DIIS enabled. MOM disabled. Fractional occupation disabled. Guess Type is SAD. Energy threshold = 1.00e-06 Density threshold = 1.00e-06 Integral threshold = 0.00e+00 ==> Primary Basis <== Basis Set: X2C-SVPALL Blend: X2C-SVPALL Number of shells: 137 Number of basis function: 309 Number of Cartesian functions: 331 Spherical Harmonics?: true Max angular momentum: 3 ==> DFT Potential <== => Composite Functional: B3LYP <= B3LYP Hyb-GGA Exchange-Correlation Functional P. J. Stephens, F. J. Devlin, C. F. Chabalowski, and M. J. Frisch, J. Phys. Chem. 98, 11623 (1994) Deriv = 1 GGA = TRUE Meta = FALSE Exchange Hybrid = TRUE MP2 Hybrid = FALSE => Exchange Functionals <= 0.0800 Slater exchange 0.7200 Becke 88 => Exact (HF) Exchange <= 0.2000 HF => Correlation Functionals <= 0.1900 Vosko, Wilk & Nusair (VWN5_RPA) 0.8100 Lee, Yang & Parr => Molecular Quadrature <= Radial Scheme = TREUTLER Pruning Scheme = FLAT Nuclear Scheme = TREUTLER BS radius alpha = 1 Pruning alpha = 1 Radial Points = 75 Spherical Points = 302 Total Points = 611550 Total Blocks = 4391 Max Points = 256 Max Functions = 271 **PSI4:PCMSOLVER Interface Active** * PCMSolver, an API for the Polarizable Continuum Model electrostatic problem. Version 1.1.11 Main authors: R. Di Remigio, L. Frediani, K. Mozgawa With contributions from: R. Bast (CMake framework) U. Ekstroem (automatic differentiation library) J. Juselius (input parsing library and CMake framework) Theory: - J. Tomasi, B. Mennucci and R. Cammi: "Quantum Mechanical Continuum Solvation Models", Chem. Rev., 105 (2005) 2999 PCMSolver is distributed under the terms of the GNU Lesser General Public License. * Git last commit hash : * Git last commit date : * Git last commit author : ~~~~~~~~~~ PCMSolver ~~~~~~~~~~ Using CODATA 2010 set of constants. Input parsing done API-side ========== Cavity Atomic radii set: Cavity type: GePol Average tesserae area = 0.3 Ang^2 Solvent probe radius = 1.385 Ang Number of spheres = 27 [initial = 27; added = 0] Number of finite elements = 1486 Number of irreducible finite elements = 1486 ============ Spheres list (in Angstrom) Sphere on Radius Alpha X Y Z -------- ---- -------- ------- ----------- ----------- ----------- 1 Rh 1.4645 1.20 0.032139 0.042136 0.007555 2 O 1.7500 1.20 0.003583 1.642407 1.211416 3 O 1.7500 1.20 1.013706 -1.019778 1.385176 4 C 1.9255 1.20 -0.040838 1.477050 2.613461 5 C 1.9255 1.20 0.732926 -0.950829 2.663617 6 C 1.9255 1.20 -0.363187 0.042141 3.051920 7 O 1.7500 1.20 1.316557 -1.609898 3.506338 8 H 1.4430 1.20 -1.323617 -0.266934 2.610532 9 H 1.4430 1.20 -0.462527 0.003978 4.145203 10 O 1.7500 1.20 -1.036231 1.037676 -1.372093 11 O 1.7500 1.20 0.026849 -1.597132 -1.185088 12 C 1.9255 1.20 -0.870980 0.856388 -2.660019 13 C 1.9255 1.20 0.165108 -1.473723 -2.585497 14 C 1.9255 1.20 0.320508 -0.019400 -3.055063 15 O 1.7500 1.20 -1.616324 1.323853 -3.502802 16 H 1.4430 1.20 1.244944 0.404437 -2.635324 17 H 1.4430 1.20 0.401586 0.005148 -4.150599 18 O 1.7500 1.20 -1.812935 -0.787160 0.353612 19 O 1.7500 1.20 1.888257 0.797614 -0.441110 20 H 1.4430 1.20 -1.594373 -1.500635 -0.301616 21 H 1.4430 1.20 -2.363512 -0.136788 -0.129257 22 H 1.4430 1.20 1.995135 1.388546 0.329373 23 H 1.4430 1.20 2.488405 0.021271 -0.352835 24 H 1.4430 1.20 -0.738991 -1.889422 -3.075362 25 H 1.4430 1.20 0.947737 1.737230 3.049933 26 H 1.4430 1.20 1.063049 -2.009692 -2.855382 27 H 1.4430 1.20 -0.830569 2.125763 2.962599 ========== Static solver Solver Type: IEFPCM, isotropic PCM matrix hermitivitized ============ Medium Medium initialized from solvent built-in data. Solvent name: Water Static permittivity = 78.39 Optical permittivity = 1.776 Solvent radius = 1.385 Ang .... Inside Green's function type: vacuum .... Outside Green's function type: uniform dielectric Permittivity = 78.39 There are 1486 tesserae, 1486 of which irreducible. => -D3MBJ: Empirical Dispersion <= Grimme's -D3 (BJ-damping, short-range refitted) Dispersion Correction Grimme S.; Ehrlich S.; Goerigk L. (2011), J. Comput. Chem., 32: 1456 Smith, D. G. A.; Burns, L. A.; Patkowski, K.; Sherrill, C. D. (2016), J. Phys. Chem. Lett.; 7: 2197 S6 = 1.000000E+00 S8 = 1.466677E+00 A1 = 2.786720E-01 A2 = 4.606311E+00 => Loading Basis Set <= Name: X2C-COULOMB-FIT Role: JKFIT Keyword: DF_BASIS_SCF atoms 1 entry RH line 2675 file /share/apps/cperry/bin/psi4/share/psi4/basis/x2c-coulomb-fit.gbs atoms 2-3, 7, 10-11, 15, 18-19 entry O line 281 file /share/apps/cperry/bin/psi4/share/psi4/basis/x2c-coulomb-fit.gbs atoms 4-6, 12-14 entry C line 181 file /share/apps/cperry/bin/psi4/share/psi4/basis/x2c-coulomb-fit.gbs atoms 8-9, 16-17, 20-27 entry H line 16 file /share/apps/cperry/bin/psi4/share/psi4/basis/x2c-coulomb-fit.gbs => Loading Basis Set <= Name: X2C-SVPALL Role: DECON Keyword: BASIS_RELATIVISTIC atoms 1 entry RH line 1583 file /share/apps/cperry/bin/psi4/share/psi4/basis/x2c-svpall.gbs atoms 2-3, 7, 10-11, 15, 18-19 entry O line 121 file /share/apps/cperry/bin/psi4/share/psi4/basis/x2c-svpall.gbs atoms 4-6, 12-14 entry C line 81 file /share/apps/cperry/bin/psi4/share/psi4/basis/x2c-svpall.gbs atoms 8-9, 16-17, 20-27 entry H line 16 file /share/apps/cperry/bin/psi4/share/psi4/basis/x2c-svpall.gbs ==> Pre-Iterations <== ------------------------------------------------------- Irrep Nso Nmo Nalpha Nbeta Ndocc Nsocc ------------------------------------------------------- A 309 309 0 0 0 0 ------------------------------------------------------- Total 309 309 79 79 79 0 ------------------------------------------------------- OEINTS: Using relativistic (X2C) overlap, kinetic, and potential integrals. ------------------------------------------------------------ Spin-Free X2C Integrals at the One-Electron Level (SFX2C-1e) by Prakash Verma and Francesco A. Evangelista ------------------------------------------------------------ ==> X2C Options <== Computational Basis: X2C-SVPALL X2C Basis: X2C-SVPALL The X2C Hamiltonian will be computed in the X2C Basis The 1-norm of |H_X2C - H_Dirac| is: 0.000000007945 ==> Integral Setup <== ==> DFJK: Density-Fitted J/K Matrices <== J tasked: Yes K tasked: Yes wK tasked: No OpenMP threads: 16 Integrals threads: 16 Memory (MB): 14648 Algorithm: Core Integral Cache: NONE Schwarz Cutoff: 1E-12 Fitting Condition: 1E-12 => Auxiliary Basis Set <= Basis Set: X2C-COULOMB-FIT Blend: X2C-COULOMB-FIT Number of shells: 439 Number of basis function: 1243 Number of Cartesian functions: 1523 Spherical Harmonics?: true Max angular momentum: 4 Minimum eigenvalue in the overlap matrix is 1.9741513485E-03. Using Symmetric Orthogonalization. SCF Guess: Superposition of Atomic Densities via on-the-fly atomic UHF. ==> Iterations <== Total Energy Delta E RMS |[F,P]| PCM polarization energy = -0.06509187746801 @DF-RKS iter 0: -5659.32878012490346 -5.65933e+03 1.01659e-01 PCM polarization energy = -0.09496942811889 @DF-RKS iter 1: -5634.69609498167756 2.46327e+01 1.11669e-02 PCM polarization energy = -0.07845832361153 @DF-RKS iter 2: -5633.89454156212832 8.01553e-01 2.22638e-02 DIIS PCM polarization energy = -0.08371243108819 @DF-RKS iter 3: -5634.92797572766904 -1.03343e+00 6.10459e-03 DIIS PCM polarization energy = -0.08487903161157 @DF-RKS iter 4: -5634.99767691672969 -6.97012e-02 1.67237e-03 DIIS PCM polarization energy = -0.08359334309402 @DF-RKS iter 5: -5635.00302466092671 -5.34774e-03 6.38242e-04 DIIS PCM polarization energy = -0.08396122918525 @DF-RKS iter 6: -5635.00396787504815 -9.43214e-04 1.57051e-04 DIIS PCM polarization energy = -0.08394603735814 @DF-RKS iter 7: -5635.00402692686112 -5.90518e-05 7.70708e-05 DIIS PCM polarization energy = -0.08394547347175 @DF-RKS iter 8: -5635.00404105133566 -1.41245e-05 2.44952e-05 DIIS PCM polarization energy = -0.08394606775589 @DF-RKS iter 9: -5635.00404336413794 -2.31280e-06 1.07547e-05 DIIS PCM polarization energy = -0.08394593511630 @DF-RKS iter 10: -5635.00404391568281 -5.51545e-07 4.04003e-06 DIIS PCM polarization energy = -0.08394473548861 @DF-RKS iter 11: -5635.00404401860214 -1.02919e-07 1.54812e-06 DIIS PCM polarization energy = -0.08394476013300 @DF-RKS iter 12: -5635.00404402762615 -9.02401e-09 1.38905e-06 DIIS PCM polarization energy = -0.08394515380190 @DF-RKS iter 13: -5635.00404403311404 -5.48789e-09 3.76794e-07 DIIS ==> Post-Iterations <== Orbital Energies (a.u.) ----------------------- Doubly Occupied: 1A -854.996946 2A -123.788774 3A -110.293835 4A -110.293521 5A -110.293374 6A -22.493082 7A -19.213813 8A -19.200014 9A -19.121253 10A -19.119104 11A -19.090617 12A -19.088301 13A -19.086001 14A -19.079686 15A -18.064032 16A -18.062286 17A -18.060014 18A -11.230413 19A -11.229155 20A -11.228769 21A -11.227635 22A -11.227178 23A -10.274770 24A -10.272514 25A -10.212705 26A -10.209281 27A -10.170654 28A -10.168587 29A -3.217744 30A -1.979143 31A -1.969707 32A -1.964010 33A -1.049975 34A -1.039272 35A -1.024633 36A -1.022734 37A -0.938772 38A -0.933434 39A -0.920897 40A -0.913469 41A -0.726283 42A -0.722727 43A -0.599882 44A -0.594512 45A -0.580403 46A -0.562421 47A -0.520665 48A -0.499222 49A -0.486172 50A -0.466215 51A -0.465468 52A -0.448890 53A -0.439533 54A -0.437648 55A -0.428316 56A -0.424916 57A -0.413166 58A -0.404606 59A -0.398737 60A -0.391018 61A -0.386520 62A -0.381419 63A -0.379907 64A -0.357466 65A -0.347652 66A -0.335781 67A -0.318981 68A -0.302573 69A -0.274370 70A -0.272228 71A -0.269177 72A -0.268026 73A -0.238776 74A -0.235590 75A -0.228142 76A -0.222410 77A -0.221615 78A -0.210324 79A -0.182651 Virtual: 80A -0.032026 81A 0.003263 82A 0.019307 83A 0.041860 84A 0.052751 85A 0.061038 86A 0.093344 87A 0.107177 88A 0.111418 89A 0.123600 90A 0.140935 91A 0.144160 92A 0.148547 93A 0.153858 94A 0.157671 95A 0.160684 96A 0.166714 97A 0.206896 98A 0.222202 99A 0.229564 100A 0.234610 101A 0.247792 102A 0.256087 103A 0.268556 104A 0.276086 105A 0.285289 106A 0.304637 107A 0.312227 108A 0.340886 109A 0.353622 110A 0.377487 111A 0.387953 112A 0.399146 113A 0.435320 114A 0.441075 115A 0.445769 116A 0.490574 117A 0.495508 118A 0.499960 119A 0.521536 120A 0.541319 121A 0.548056 122A 0.554373 123A 0.570134 124A 0.576140 125A 0.587756 126A 0.595777 127A 0.604986 128A 0.609738 129A 0.635145 130A 0.642379 131A 0.653852 132A 0.662402 133A 0.664832 134A 0.672229 135A 0.680533 136A 0.683892 137A 0.686596 138A 0.697455 139A 0.700368 140A 0.702424 141A 0.703845 142A 0.717133 143A 0.738714 144A 0.760205 145A 0.775418 146A 0.810456 147A 0.833144 148A 0.849191 149A 0.869447 150A 0.871155 151A 0.884795 152A 0.915865 153A 0.926290 154A 0.933036 155A 0.947522 156A 0.968947 157A 0.975024 158A 1.012627 159A 1.027849 160A 1.041947 161A 1.052161 162A 1.089064 163A 1.093360 164A 1.111289 165A 1.119993 166A 1.141502 167A 1.151358 168A 1.172983 169A 1.183707 170A 1.186544 171A 1.219619 172A 1.245118 173A 1.255009 174A 1.262473 175A 1.276894 176A 1.290172 177A 1.309387 178A 1.315753 179A 1.327700 180A 1.342156 181A 1.367995 182A 1.398383 183A 1.409528 184A 1.423637 185A 1.443915 186A 1.454686 187A 1.465590 188A 1.472055 189A 1.508070 190A 1.519879 191A 1.536830 192A 1.545517 193A 1.566815 194A 1.583640 195A 1.612066 196A 1.622982 197A 1.663806 198A 1.673814 199A 1.682203 200A 1.691456 201A 1.703308 202A 1.724901 203A 1.729632 204A 1.740190 205A 1.760395 206A 1.773286 207A 1.777318 208A 1.793865 209A 1.798373 210A 1.806282 211A 1.811932 212A 1.817659 213A 1.838801 214A 1.850286 215A 1.859241 216A 1.868491 217A 1.889026 218A 1.891332 219A 1.900930 220A 1.904814 221A 1.924443 222A 1.940120 223A 1.949793 224A 1.961632 225A 1.989861 226A 2.008815 227A 2.034242 228A 2.038022 229A 2.067274 230A 2.099185 231A 2.111723 232A 2.126014 233A 2.138117 234A 2.142524 235A 2.157253 236A 2.167016 237A 2.175563 238A 2.182479 239A 2.193612 240A 2.226664 241A 2.254298 242A 2.265792 243A 2.291905 244A 2.314358 245A 2.321297 246A 2.338229 247A 2.385364 248A 2.400964 249A 2.432659 250A 2.473478 251A 2.482013 252A 2.490497 253A 2.499793 254A 2.549941 255A 2.595957 256A 2.601431 257A 2.675879 258A 2.687991 259A 2.693783 260A 2.707124 261A 2.740916 262A 2.741386 263A 2.766573 264A 2.794292 265A 2.813133 266A 2.825215 267A 2.835924 268A 2.847783 269A 2.858245 270A 2.876548 271A 2.892626 272A 2.918719 273A 2.948632 274A 2.958189 275A 2.963916 276A 2.976998 277A 2.981834 278A 3.008211 279A 3.023152 280A 3.030262 281A 3.041220 282A 3.062293 283A 3.090409 284A 3.119569 285A 3.125239 286A 3.143710 287A 3.150744 288A 3.174211 289A 3.198107 290A 3.244466 291A 3.248166 292A 3.299221 293A 3.385611 294A 3.435496 295A 3.454144 296A 3.459704 297A 3.489523 298A 3.525216 299A 3.636530 300A 3.639197 301A 3.674372 302A 3.689819 303A 3.803386 304A 3.817984 305A 4.114563 306A 4.174568 307A 4.248175 308A 4.446264 309A 4.504195 Final Occupation by Irrep: A DOCC [ 79 ] Energy converged. @DF-RKS Final Energy: -5635.00404403311404 => Energetics <= Nuclear Repulsion Energy = 2018.5982855052438936 One-Electron Energy = -11807.4021601024014672 Two-Electron Energy = 4352.3276754430444271 DFT Exchange-Correlation Energy = -198.3734912351995376 Empirical Dispersion Energy = -0.0704084900000000 VV10 Nonlocal Energy = 0.0000000000000000 PCM Polarization Energy = -0.0839451538019041 Total Energy = -5635.0040440331140417 Alert: EFP and PCM quantities not currently incorporated into SCF psivars. Properties will be evaluated at 0.000000, 0.000000, 0.000000 Bohr Properties computed using the SCF density matrix Nuclear Dipole Moment: (a.u.) X: 0.3826 Y: -0.6246 Z: -0.4699 Electronic Dipole Moment: (a.u.) X: 0.2862 Y: 0.6848 Z: 0.1558 Dipole Moment: (a.u.) X: 0.6688 Y: 0.0602 Z: -0.3141 Total: 0.7413 Dipole Moment: (Debye) X: 1.6999 Y: 0.1531 Z: -0.7983 Total: 1.8843 *** tstop() called on compute-0-27.local at Mon Jan 29 08:40:49 2018 Module time: user time = 1729.77 seconds = 28.83 minutes system time = 4.15 seconds = 0.07 minutes total time = 1271 seconds = 21.18 minutes Total time: user time = 1729.77 seconds = 28.83 minutes system time = 4.15 seconds = 0.07 minutes total time = 1271 seconds = 21.18 minutes *** Psi4 exiting successfully. Buy a developer a beer!
It is true PCMSolver should only contribute to a fraction of the cost, but it can become a bottleneck in parallel calculations (Amdahl’s law). In the calculation reported here the nr. of finite elements (1400) is basically as large as the nr. of basis functions (1200), making the PCM integrals in practice a three-index quantity, rather than two.
The first step would be to parallelize integral calculations. For truly big systems one should also consider screening methods for the integrals (many PCM integrals can and should be converted to point-charge potentials) and FMM methods for the apparent surface problem. Another quite well known 
code already has these features.
I agree to that using the pcmsolver in psi4 for the scf iteration part more or less only use one thread (when looking at process cpu usage in top, it stays at 100% most of the time).
Are there any plans to make this type of calculation scale better?
A simple acetic acid calculation on 4 cores (pbe/def2-kttvppd)
pcm:
user time = 217.58 seconds = 3.63 minutes
system time = 1.13 seconds = 0.02 minutes
total time = 163 seconds = 2.72 minutes
no pcm:
user time = 72.72 seconds = 1.21 minutes
system time = 0.52 seconds = 0.01 minutes
total time = 21 seconds = 0.35 minutes
Regards
Termo
We should look into this, depending on where the PCM bottleneck is it should be a fairly straightforward parallelization.
I would very much like to help on this. Currently I have set pcmsolver up with git and I’m trying to see if I can get the openMP part to work in pcmsolver that have been pushed recently.
please let me know if I can be of any assistance!
That would be great @robertodr, do you have any suggested starting points?
If you could make an issue on github.com/psi4/psi4 we can getting the broader development community to pitch in ideas.
Is there any progress on parallelisation of PCMSolver? I just run few calculations using PCMsolver and it is mostly running as a single core process.
There is! The PR to watch is https://github.com/psi4/psi4/pull/2388. It will soon be merged.