Merge pull request #485 from toxa81/develop

update travis ci and front page documentation

.travis.yml

@@ -16,7 +16,7 @@ branches:
 #    - develop
 
 env:
-    - CC=mpicc CXX=mpic++ FC=mpif90 FCCPP=cpp SIRIUS_BINARIES=/home/travis/local/bin PATH=/home/travis/local/bin:$PATH
+    - CC=mpicc CXX=mpic++ FC=mpif90 FCCPP=cpp PATH=/home/travis/local/bin:$PATH
 
 before_install:
     - ./check_format.x
@@ -27,6 +27,7 @@ install:
     - ./ci/travis_build_sirius.sh
 
 script:
+    - ./ci/travis_run_unit_tests.sh
     - ./ci/travis_run_verification.sh
 
 #after_success:

README.md

@@ -21,6 +21,8 @@
 * [Accelerating DFT codes](#accelerating-dft-codes)
   * [Quantum ESPRESSO](#quantum-espresso)
 * [Examples](#examples)
+* [Contacts](#contacts)
+* [Acknowledgements](#acknowledgements)
 
 ## Introduction
 SIRIUS is a domain specific library for electronic structure calculations. It implements pseudopotential plane wave (PP-PW)
@@ -305,16 +307,57 @@ SIRIUS library is usgin OpenMP for node-level parallelization. To run QE/SIRIUS
  * by default, use one rank per node and many OMP threads; if the calculated system is really small, try to saturate 
    the GPU card using more MPI ranks (e.g.: on a 12-core node, use 2-3-4 ranks with 6-4-3 OMP threads)
 
-#### Example: ground state of Si511Ge
-In the following example we compare performace of native and SIRIUS-enabled versions of QE. Native QE was run on the
-dual-socket nodes containing two 18-core Intel Haswell CPUs.
+### Benchmarks
+In the following examples we compare performace of native and SIRIUS-enabled versions of QE. CPU-only runs were executed
+on the dual-socket multi-core nodes containing two 18-core Intel Broadwell CPUs. GPU rus were executed on the hybrid
+nodes containing 12-core Intel Haswell CPU and NVIDIA Tesla P100 card:
 
+|Hybrid partition (Cray XC50)                | Multicore partition (Cray XC40)                  |
+|--------------------------------------------|--------------------------------------------------|
+|Intel Xeon E5-2690 v3 @2.60GHz, 12 cores <br> NVIDIA Tesla P100 16GB | Two Intel Xeon E5-2695 v4 @2.10GHz (2 x 18 cores)|
+
+Ground state calculation ([input](https://github.com/electronic-structure/benchmarks/tree/master/performance/Si511Ge))
+of Si511Ge.
 
 <p align="center">
 <img src="doc/images/Si511Ge_perf.png">
 </p>
 
 
 
-
 ## Examples
+
+## Contacts
+Have you got any questions, feel free to contact us:
+  * Anton Kozhevnikov ([email protected])
+  * Mathieu Taillefumier ([email protected])
+  * Simon Pintarelli ([email protected])
+
+## Acknowledgements
+The development of SIRIUS library would not be possible without support of the following organizations:
+
+|List of supporting organizations                  |
+|:------------------------------------------------:|
+|                                                  |
+| Swiss Federal Institute of Technology in Zürich  |                                                  |
+| ![ethz](doc/images/logo_ethz.png)                |
+| https://www.ethz.ch/                             |
+|                                                  |
+| Swiss National Supercomputing Centre             |
+| ![cscs](doc/images/logo_cscs.png)                |
+|  https://www.cscs.ch/                            |
+|                                                  |
+|  Platform for Advanced Scientific Computing      |
+|  ![pasc](doc/images/logo_pasc.png)               |
+|  https://www.pasc-ch.org/                        |
+|                                                  |
+|  MAX (MAterials design at the eXascale) <br> European Centre of Excellence |
+| ![pasc](doc/images/logo_max.png)                                           |
+|  http://www.max-centre.eu/                                                 |
+|                                                  |
+|  Partnership for Advanced Computing in Europe    |
+| ![pasc](doc/images/logo_prace.png)               |
+|  https://prace-ri.eu/                            |
+
+
+

ci/travis_build_sirius.sh

@@ -2,6 +2,6 @@
 
 mkdir build
 cd build
-cmake ../ -DCMAKE_BUILD_TYPE=RelWithDebInfo -DSpFFT_DIR=/home/travis/local/lib/cmake/SpFFT -DCMAKE_INSTALL_PREFIX=/home/travis/local
+cmake ../ -DCMAKE_BUILD_TYPE=RelWithDebInfo -DSpFFT_DIR=/home/travis/local/lib/cmake/SpFFT -DCMAKE_INSTALL_PREFIX=/home/travis/local -DBUILD_TESTS=1
 make -j 2 install
 

ci/travis_run_unit_tests.sh

@@ -0,0 +1,3 @@
+#!/bin/bash
+cd build/apps/unit_tests
+./unit_tests.x

src/api/generated.f90

@@ -427,9 +427,13 @@ end subroutine sirius_set_parameters
 !> @param [out] iter_solver_tol_empty Tolerance for the empty states.
 !> @param [out] verbosity Verbosity level.
 !> @param [out] hubbard_correction True if LDA+U correction is enabled.
+!> @param [out] evp_work_count Internal counter of total eigen-value problem work.
+!> @param [out] num_loc_op_applied Internal counter of the number of wave-functions to which Hamiltonian was applied.
+!> @param [out] error_code Error code.
 subroutine sirius_get_parameters(handler,lmax_apw,lmax_rho,lmax_pot,num_fv_states,&
 &num_bands,num_mag_dims,pw_cutoff,gk_cutoff,fft_grid_size,auto_rmt,gamma_point,use_symmetry,&
-&so_correction,iter_solver_tol,iter_solver_tol_empty,verbosity,hubbard_correction)
+&so_correction,iter_solver_tol,iter_solver_tol_empty,verbosity,hubbard_correction,&
+&evp_work_count,num_loc_op_applied,error_code)
 implicit none
 type(C_PTR), intent(in) :: handler
 integer(C_INT), optional, target, intent(out) :: lmax_apw
@@ -449,6 +453,9 @@ subroutine sirius_get_parameters(handler,lmax_apw,lmax_rho,lmax_pot,num_fv_state
 real(C_DOUBLE), optional, target, intent(out) :: iter_solver_tol_empty
 integer(C_INT), optional, target, intent(out) :: verbosity
 logical(C_BOOL), optional, target, intent(out) :: hubbard_correction
+real(C_DOUBLE), optional, target, intent(out) :: evp_work_count
+integer(C_INT), optional, target, intent(out) :: num_loc_op_applied
+integer(C_INT), optional, target, intent(out) :: error_code
 type(C_PTR) :: lmax_apw_ptr
 type(C_PTR) :: lmax_rho_ptr
 type(C_PTR) :: lmax_pot_ptr
@@ -466,10 +473,14 @@ subroutine sirius_get_parameters(handler,lmax_apw,lmax_rho,lmax_pot,num_fv_state
 type(C_PTR) :: iter_solver_tol_empty_ptr
 type(C_PTR) :: verbosity_ptr
 type(C_PTR) :: hubbard_correction_ptr
+type(C_PTR) :: evp_work_count_ptr
+type(C_PTR) :: num_loc_op_applied_ptr
+type(C_PTR) :: error_code_ptr
 interface
 subroutine sirius_get_parameters_aux(handler,lmax_apw,lmax_rho,lmax_pot,num_fv_states,&
 &num_bands,num_mag_dims,pw_cutoff,gk_cutoff,fft_grid_size,auto_rmt,gamma_point,use_symmetry,&
-&so_correction,iter_solver_tol,iter_solver_tol_empty,verbosity,hubbard_correction)&
+&so_correction,iter_solver_tol,iter_solver_tol_empty,verbosity,hubbard_correction,&
+&evp_work_count,num_loc_op_applied,error_code)&
 &bind(C, name="sirius_get_parameters")
 use, intrinsic :: ISO_C_BINDING
 type(C_PTR), intent(in) :: handler
@@ -490,6 +501,9 @@ subroutine sirius_get_parameters_aux(handler,lmax_apw,lmax_rho,lmax_pot,num_fv_s
 type(C_PTR), value :: iter_solver_tol_empty
 type(C_PTR), value :: verbosity
 type(C_PTR), value :: hubbard_correction
+type(C_PTR), value :: evp_work_count
+type(C_PTR), value :: num_loc_op_applied
+type(C_PTR), value :: error_code
 end subroutine
 end interface
 
@@ -544,10 +558,19 @@ subroutine sirius_get_parameters_aux(handler,lmax_apw,lmax_rho,lmax_pot,num_fv_s
 hubbard_correction_ptr = C_NULL_PTR
 if (present(hubbard_correction)) hubbard_correction_ptr = C_LOC(hubbard_correction)
 
+evp_work_count_ptr = C_NULL_PTR
+if (present(evp_work_count)) evp_work_count_ptr = C_LOC(evp_work_count)
+
+num_loc_op_applied_ptr = C_NULL_PTR
+if (present(num_loc_op_applied)) num_loc_op_applied_ptr = C_LOC(num_loc_op_applied)
+
+error_code_ptr = C_NULL_PTR
+if (present(error_code)) error_code_ptr = C_LOC(error_code)
+
 call sirius_get_parameters_aux(handler,lmax_apw_ptr,lmax_rho_ptr,lmax_pot_ptr,num_fv_states_ptr,&
 &num_bands_ptr,num_mag_dims_ptr,pw_cutoff_ptr,gk_cutoff_ptr,fft_grid_size_ptr,auto_rmt_ptr,&
 &gamma_point_ptr,use_symmetry_ptr,so_correction_ptr,iter_solver_tol_ptr,iter_solver_tol_empty_ptr,&
-&verbosity_ptr,hubbard_correction_ptr)
+&verbosity_ptr,hubbard_correction_ptr,evp_work_count_ptr,num_loc_op_applied_ptr,error_code_ptr)
 end subroutine sirius_get_parameters
 
 !> @brief Add one of the XC functionals.

src/api/sirius_api.cpp

@@ -415,6 +415,9 @@ void sirius_set_parameters(void*  const* handler__,
    @fortran argument out optional double iter_solver_tol_empty    Tolerance for the empty states.
    @fortran argument out optional int    verbosity                Verbosity level.
    @fortran argument out optional bool   hubbard_correction       True if LDA+U correction is enabled.
+   @fortran argument out optional double evp_work_count           Internal counter of total eigen-value problem work.
+   @fortran argument out optional int    num_loc_op_applied       Internal counter of the number of wave-functions to which Hamiltonian was applied.
+   @fortran argument out optional int    error_code               Error code.
    @fortran end */
 void sirius_get_parameters(void* const* handler__,
                            int*         lmax_apw__,
@@ -433,62 +436,74 @@ void sirius_get_parameters(void* const* handler__,
                            double*      iter_solver_tol__,
                            double*      iter_solver_tol_empty__,
                            int*         verbosity__,
-                           bool*        hubbard_correction__)
+                           bool*        hubbard_correction__,
+                           double*      evp_work_count__,
+                           int*         num_loc_op_applied__,
+                           int*         error_code__)
 {
-    auto& sim_ctx = get_sim_ctx(handler__);
-    if (lmax_apw__ != nullptr) {
-        *lmax_apw__ = sim_ctx.lmax_apw();
-    }
-    if (lmax_rho__ != nullptr) {
-        *lmax_rho__ = sim_ctx.lmax_rho();
-    }
-    if (lmax_pot__ != nullptr) {
-        *lmax_pot__ = sim_ctx.lmax_pot();
-    }
-    if (num_fv_states__ != nullptr) {
-        *num_fv_states__ = sim_ctx.num_fv_states();
-    }
-    if (num_bands__ != nullptr) {
-        *num_bands__ = sim_ctx.num_bands();
-    }
-    if (num_mag_dims__ != nullptr) {
-        *num_mag_dims__ = sim_ctx.num_mag_dims();
-    }
-    if (pw_cutoff__ != nullptr) {
-        *pw_cutoff__ = sim_ctx.pw_cutoff();
-    }
-    if (gk_cutoff__ != nullptr) {
-        *gk_cutoff__ = sim_ctx.gk_cutoff();
-    }
-    if (auto_rmt__ != nullptr) {
-        *auto_rmt__ = sim_ctx.auto_rmt();
-    }
-    if (gamma_point__ != nullptr) {
-        *gamma_point__ = sim_ctx.gamma_point();
-    }
-    if (use_symmetry__ != nullptr) {
-        *use_symmetry__ = sim_ctx.use_symmetry();
-    }
-    if (so_correction__ != nullptr) {
-        *so_correction__ = sim_ctx.so_correction();
-    }
-    if (iter_solver_tol__ != nullptr) {
-        *iter_solver_tol__ = sim_ctx.iterative_solver_tolerance();
-    }
-    if (iter_solver_tol_empty__ != nullptr) {
-        *iter_solver_tol_empty__ = sim_ctx.iterative_solver_input().empty_states_tolerance_;
-    }
-    if (verbosity__ != nullptr) {
-        *verbosity__ = sim_ctx.control().verbosity_;
-    }
-    if (hubbard_correction__ != nullptr) {
-        *hubbard_correction__ = sim_ctx.hubbard_correction();
-    }
-    if (fft_grid_size__ != nullptr) {
-        for (int x: {0, 1, 2}) {
-            fft_grid_size__[x] = sim_ctx.fft_grid()[x];
+    call_sirius([&]()
+    {
+        auto& sim_ctx = get_sim_ctx(handler__);
+        if (lmax_apw__ != nullptr) {
+            *lmax_apw__ = sim_ctx.lmax_apw();
         }
-    }
+        if (lmax_rho__ != nullptr) {
+            *lmax_rho__ = sim_ctx.lmax_rho();
+        }
+        if (lmax_pot__ != nullptr) {
+            *lmax_pot__ = sim_ctx.lmax_pot();
+        }
+        if (num_fv_states__ != nullptr) {
+            *num_fv_states__ = sim_ctx.num_fv_states();
+        }
+        if (num_bands__ != nullptr) {
+            *num_bands__ = sim_ctx.num_bands();
+        }
+        if (num_mag_dims__ != nullptr) {
+            *num_mag_dims__ = sim_ctx.num_mag_dims();
+        }
+        if (pw_cutoff__ != nullptr) {
+            *pw_cutoff__ = sim_ctx.pw_cutoff();
+        }
+        if (gk_cutoff__ != nullptr) {
+            *gk_cutoff__ = sim_ctx.gk_cutoff();
+        }
+        if (auto_rmt__ != nullptr) {
+            *auto_rmt__ = sim_ctx.auto_rmt();
+        }
+        if (gamma_point__ != nullptr) {
+            *gamma_point__ = sim_ctx.gamma_point();
+        }
+        if (use_symmetry__ != nullptr) {
+            *use_symmetry__ = sim_ctx.use_symmetry();
+        }
+        if (so_correction__ != nullptr) {
+            *so_correction__ = sim_ctx.so_correction();
+        }
+        if (iter_solver_tol__ != nullptr) {
+            *iter_solver_tol__ = sim_ctx.iterative_solver_tolerance();
+        }
+        if (iter_solver_tol_empty__ != nullptr) {
+            *iter_solver_tol_empty__ = sim_ctx.iterative_solver_input().empty_states_tolerance_;
+        }
+        if (verbosity__ != nullptr) {
+            *verbosity__ = sim_ctx.control().verbosity_;
+        }
+        if (hubbard_correction__ != nullptr) {
+            *hubbard_correction__ = sim_ctx.hubbard_correction();
+        }
+        if (fft_grid_size__ != nullptr) {
+            for (int x: {0, 1, 2}) {
+                fft_grid_size__[x] = sim_ctx.fft_grid()[x];
+            }
+        }
+        if (evp_work_count__ != nullptr) {
+            *evp_work_count__ = sim_ctx.evp_work_count();
+        }
+        if (num_loc_op_applied__ != nullptr) {
+            *num_loc_op_applied__ = sim_ctx.num_loc_op_applied();
+        }
+    }, error_code__);
 }
 
 
@@ -3581,5 +3596,4 @@ void sirius_set_callback_function(void* const* handler__, char const* label__, v
     }, error_code__);
 }
 
-
 } // extern "C"

src/mpi/communicator.hpp

@@ -412,6 +412,14 @@ class Communicator
         return (j__ * (j__ + 1) / 2 + i__ + 1) << 6;
     }
 
+    static std::string processor_name()
+    {
+        char name[MPI_MAX_PROCESSOR_NAME];
+        int len;
+        CALL_MPI(MPI_Get_processor_name, (name, &len));
+        return std::string(name, len);
+    }
+
     /// Rank of MPI process inside communicator.
     inline int rank() const
     {

src/simulation_context.cpp

@@ -27,6 +27,7 @@
 #include "simulation_context.hpp"
 #include "symmetry/find_lat_sym.hpp"
 #include "utils/profiler.hpp"
+#include "utils/env.hpp"
 
 namespace sirius {
 
@@ -280,14 +281,17 @@ void Simulation_context::initialize()
     /* initialize MPI communicators */
     init_comm();
 
-    if (control().verbosity_ >= 3) {
+    auto print_mpi_layout = utils::get_env<int>("SIRIUS_PRINT_MPI_LAYOUT");
+
+    if (control().verbosity_ >= 3 || (print_mpi_layout && *print_mpi_layout)) {
         pstdout pout(comm());
         if (comm().rank() == 0) {
             pout.printf("MPI rank placement\n");
             pout.printf("------------------\n");
         }
-        pout.printf("rank: %3i, comm_band_rank: %3i, comm_k_rank: %3i, hostname: %s\n", comm().rank(),
-                    comm_band().rank(), comm_k().rank(), utils::hostname().c_str());
+        auto name = sddk::Communicator::processor_name();
+        pout.printf("rank: %3i, comm_band_rank: %3i, comm_k_rank: %3i, hostname: %s, mpi processor name: %s\n",
+            comm().rank(), comm_band().rank(), comm_k().rank(), utils::hostname().c_str(), name.c_str());
     }
 
     switch (processing_unit()) {