Parallelism and static

plugins/Full_CI_ZMQ/pt2_stoch_routines.irp.f

@@ -204,7 +204,7 @@ subroutine ZMQ_pt2(E, pt2,relative_error, error)
         nproc_target = min(nproc_target,nproc)
       endif
 
-      call omp_set_nested(.true.)
+      call omp_set_nested(.false.)
 
       !$OMP PARALLEL DEFAULT(shared) NUM_THREADS(nproc_target+1)            &
           !$OMP  PRIVATE(i)

plugins/Full_CI_ZMQ/selection_davidson_slave.irp.f

@@ -4,6 +4,7 @@ program selection_slave
 ! Helper program to compute the PT2 in distributed mode.
   END_DOC
 
+  call omp_set_nested(.false.)
   read_wf = .False.
   distributed_davidson = .False.
   SOFT_TOUCH read_wf distributed_davidson

plugins/dress_zmq/dress_stoch_routines.irp.f

@@ -16,7 +16,8 @@ END_PROVIDER
   integer :: e
   e = elec_num - n_core_orb * 2
   pt2_n_tasks_max = 1 + min((e*(e-1))/2, int(dsqrt(dble(N_det_generators)))/10)
-  do i=1,N_det_generators
+  pt2_F(:) = 1
+  do i=1,min(10000,N_det_generators)
     pt2_F(i) = 1 + dble(pt2_n_tasks_max)*maxval(dsqrt(dabs(psi_coef_sorted_gen(i,1:N_states))))
   enddo
 
@@ -25,7 +26,7 @@ END_PROVIDER
     pt2_N_teeth = 1
   else
     pt2_minDetInFirstTeeth = min(5, N_det_generators)
-    do pt2_N_teeth=100,2,-1
+    do pt2_N_teeth=50,2,-1
       if(testTeethBuilding(pt2_minDetInFirstTeeth, pt2_N_teeth)) exit
     end do
   end if
@@ -127,7 +128,7 @@ END_PROVIDER
 &BEGIN_PROVIDER [ integer, dress_N_cp ]
   implicit none
   integer :: N_c, N_j, U, t, i, m
-  double precision :: v
+  double precision :: v, dt
   double precision, allocatable :: tilde_M(:)
   logical, allocatable :: d(:)
   integer, external :: dress_find_sample
@@ -172,11 +173,13 @@ END_PROVIDER
   U = 0
   
   m = 1
+  ! TODO Slow loop : to optimize
   do while(N_j < N_det_generators)
     !ADD_COMB
     N_c += 1
+    dt = 0.d0
     do t=0, pt2_N_teeth-1
-      v = pt2_u_0 + pt2_W_T * (dble(t) + pt2_u(N_c))
+      v = pt2_u_0 + pt2_W_T * (dt + pt2_u(N_c))
       i = dress_find_sample(v, pt2_cW)
       tilde_M(i) += 1d0
       if(.not. d(i)) then
@@ -184,6 +187,7 @@ END_PROVIDER
         pt2_J_(N_j) = i
         d(i) = .true.
       end if
+      dt = dt + 1.d0
     end do
    
     !FILL_TOOTH
@@ -193,7 +197,7 @@ END_PROVIDER
         N_j += 1
         pt2_J_(N_j) = U
         d(U) = .true.
-        exit;
+        exit
       end if
     end do
     
@@ -254,7 +258,7 @@ subroutine ZMQ_dress(E, dress, delta_out, delta_s2_out, relative_error)
     state_average_weight(dress_stoch_istate) = 1.d0
     TOUCH state_average_weight dress_stoch_istate
     
-    provide nproc mo_bielec_integrals_in_map mo_mono_elec_integral psi_selectors pt2_F
+    provide nproc mo_bielec_integrals_in_map mo_mono_elec_integral psi_selectors pt2_F pt2_N_teeth dress_M_m
     
     print *, '========== ================= ================= ================='
     print *, ' Samples        Energy         Stat. Error         Seconds      '
@@ -350,8 +354,6 @@ subroutine ZMQ_dress(E, dress, delta_out, delta_s2_out, relative_error)
       nproc_target = min(nproc_target,nproc)
     endif
 
-    call omp_set_nested(.true.)
-
     !$OMP PARALLEL DEFAULT(shared) NUM_THREADS(2)              &
         !$OMP  PRIVATE(i)
     i = omp_get_thread_num()
@@ -372,7 +374,6 @@ subroutine ZMQ_dress(E, dress, delta_out, delta_s2_out, relative_error)
   enddo
   FREE dress_stoch_istate
   state_average_weight(:) = state_average_weight_save(:)
-!    call omp_set_nested(.false.)
   TOUCH state_average_weight
   deallocate(delta,delta_s2)
   
@@ -631,7 +632,7 @@ integer function dress_find_sample(v, w)
   r = N_det_generators
 
   do while(r-l > 1)
-    i = (r+l) / 2
+    i = ishft(r+l,-1)
     if(w(i) < v) then
       l = i
     else

plugins/dress_zmq/run_dress_slave.irp.f

@@ -43,6 +43,7 @@ subroutine run_dress_slave(thread,iproce,energy)
   integer :: ntask_tbd, task_tbd(Nproc), i_gen_tbd(Nproc), subset_tbd(Nproc)
   logical :: interesting
 
+  PROVIDE dress_dot_F psi_coef dress_stoch_istate dress_e N_int 
 
   allocate(delta_det(N_states, N_det, 0:pt2_N_teeth+1, 2))
   allocate(cp(N_states, N_det, dress_N_cp, 2))
@@ -71,6 +72,8 @@ subroutine run_dress_slave(thread,iproce,energy)
   provide psi_energy
   ending = dress_N_cp+1
   ntask_tbd = 0
+  call omp_set_nested(.true.)
+
   !$OMP PARALLEL DEFAULT(SHARED) &
   !$OMP PRIVATE(interesting, breve_delta_m, task_id) &
   !$OMP PRIVATE(tmp,fac,m,l,t,sum_f,n_tasks) &
@@ -81,6 +84,7 @@ subroutine run_dress_slave(thread,iproce,energy)
   zmq_socket_push      = new_zmq_push_socket(thread)
   integer, external :: connect_to_taskserver
   !$OMP CRITICAL
+  call omp_set_nested(.false.)
   if (connect_to_taskserver(zmq_to_qp_run_socket,worker_id,thread) == -1) then
     print *,  irp_here, ': Unable to connect to task server'
     stop -1
@@ -292,6 +296,7 @@ subroutine run_dress_slave(thread,iproce,energy)
   !$OMP END CRITICAL
 
   !$OMP END PARALLEL
+  call omp_set_nested(.false.)
 !  do i=0,dress_N_cp+1
 !    call omp_destroy_lock(lck_sto(i))
 !  end do

plugins/shiftedbk/shifted_bk.irp.f

@@ -8,6 +8,7 @@ program shifted_bk
   PROVIDE psi_bilinear_matrix_rows psi_det_sorted_gen_order psi_bilinear_matrix_order
   PROVIDE psi_bilinear_matrix_transp_rows_loc psi_bilinear_matrix_transp_columns
   PROVIDE psi_bilinear_matrix_transp_order
+  PROVIDE psi_energy
 
 
   !call diagonalize_CI()

plugins/shiftedbk/shifted_bk_slave.irp.f

@@ -151,10 +151,8 @@ subroutine run_w
       IRP_ENDIF
 
       if (.true.) then
-        !$OMP PARALLEL PRIVATE(i)
+        call omp_set_nested(.True.)
-        i = omp_get_thread_num()
         call run_dress_slave(0,i,dress_e0_denominator)
-        !$OMP END PARALLEL
       endif
       print *,  'PT2 done'
       FREE state_average_weight

scripts/make_binary.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/bin/bash -x
 # 
 # Creates a self-contained binary distribution in the form of a tar.gz file
 # 
@@ -22,7 +22,7 @@ if [[ -f quantum_package.rc \
    && -d ocaml \
    && -d scripts ]]
 then
-  head -1 README.md | grep "Quantum package" > /dev/null
+  head -1 README.md | grep "IMPORTANT" > /dev/null
   if [[ $? -ne 0 ]]
   then
     echo "This doesn't look like a quantum_package directory  (README.md)"
@@ -35,14 +35,11 @@ fi
 
 
 # Build all sources
-for dir in ${QP_ROOT}/{src,ocaml}
+#ninja
-do
+#if [[ $? -ne 0 ]]
-  make -C ${dir}
+#then
-  if [[ $? -ne 0 ]]
+#  echo "Error building ${dir}"
-  then
+#fi
-    echo "Error building ${dir}"
-  fi
-done
 
 
 # Copy the files in the static directory
@@ -68,10 +65,10 @@ echo "Creating root of static directory"
 #     ---------------------------------
 
 rm -rf -- ${QPACKAGE_STATIC}
-mkdir -p -- ${QPACKAGE_STATIC}/{bin,lib,extra_lib,data}
+mkdir -p -- ${QPACKAGE_STATIC}/{bin,lib,extra_lib,data,install}
 if [[ $? -ne 0 ]] ;
 then
-  echo "Error creating ${QPACKAGE_STATIC}/{bin,lib,extra_lib,data}"
+  echo "Error creating ${QPACKAGE_STATIC}/{bin,lib,extra_lib,data,install}"
   exit 1
 fi
 
@@ -97,7 +94,7 @@ fi
 cp -- ${FORTRAN_EXEC} ${OCAML_EXEC} ${QPACKAGE_STATIC}/bin
 if [[ $? -ne 0 ]] ;
 then
-  echo 'cp -- ${FORTRAN_EXEC} ${OCAML_EXEC} ${QPACKAGE_STATIC}/bin'
+  cp -- ${FORTRAN_EXEC} ${OCAML_EXEC} ${QPACKAGE_STATIC}/bin
   exit 1
 fi
 
@@ -143,10 +140,10 @@ cp -- ${QPACKAGE_STATIC}/extra_lib/lib{[gi]omp*,mkl*,lapack*,blas*,z*} ${QPACKAG
 echo "Copying EMSL_Basis directory"
 #     ---------------------------- 
 
-cp -r -- ${QP_ROOT}/EMSL_Basis ${QPACKAGE_STATIC}/
+cp -r -- ${QP_ROOT}/install/emsl ${QPACKAGE_STATIC}/install
 if [[ $? -ne 0 ]] ;
 then
-  echo 'cp -r -- ${QP_ROOT}/EMSL_Basis ${QPACKAGE_STATIC}/'
+  echo 'cp -r -- ${QP_ROOT}/install/emsl ${QPACKAGE_STATIC}/'
   exit 1
 fi
 
@@ -169,12 +166,27 @@ echo "Creating quantum_package.rc"
 
 cat << EOF > ${QPACKAGE_STATIC}/quantum_package.rc
 export QP_ROOT=\$( cd \$(dirname "\${BASH_SOURCE}")  ; pwd -P )
-export LD_LIBRARY_PATH="\${QP_ROOT}"/lib:\${LD_LIBRARY_PATH}
+
-export LIBRARY_PATH="\${QP_ROOT}"/lib:\${LIBRARY_PATH}
+export QP_EZFIO=\${QP_ROOT}/install/EZFIO
-export PYTHONPATH="\${QP_ROOT}"/scripts:\${PYTHONPATH}
+export QP_PYTHON=\${QP_ROOT}/scripts:\${QP_ROOT}/scripts/ezfio_interface:\${QP_ROOT}/scripts/utility:\${QP_ROOT}/scripts/module:\${QP_ROOT}/scripts/pseudo:\${QP_ROOT}/scripts/compilation:\${QP_ROOT}/install/bats:\${QP_ROOT}/install/Downloads:\${QP_ROOT}/install/eigen:\${QP_ROOT}/install/p_graphviz:\${QP_ROOT}/install/gmp:\${QP_ROOT}/install/resultsFile:\${QP_ROOT}/install/_build:\${QP_ROOT}/install/emsl:\${QP_ROOT}/install/scripts:\${QP_ROOT}/install/docopt:\${QP_ROOT}/install/irpf90:\${QP_ROOT}/install/zlib:\${QP_ROOT}/install/EZFIO
-export PATH="\${QP_ROOT}"/scripts:\${PATH}
+
-export PATH="\${QP_ROOT}"/bin:\${PATH}
+export IRPF90=\${QP_ROOT}/bin/irpf90
-export PATH="\${QP_ROOT}"/ocaml:\${PATH}
+export NINJA=\${QP_ROOT}/bin/ninja
+function qp_prepend_export () {
+eval "value_1="\\\${\$1}""
+if [[ -z \$value_1 ]] ; then
+    echo "\${2}:"
+else
+    echo "\${2}:\${value_1}"
+fi
+}
+export PYTHONPATH=\$(qp_prepend_export "PYTHONPATH" "\${QP_EZFIO}/Python":"\${QP_PYTHON}")
+export PATH=\$(qp_prepend_export "PATH" "\${QP_PYTHON}":"\${QP_ROOT}"/bin:"\${QP_ROOT}"/ocaml)
+export LD_LIBRARY_PATH=\$(qp_prepend_export "LD_LIBRARY_PATH" "\${QP_ROOT}"/lib:"\${QP_ROOT}"/extra_lib:"\${QP_ROOT}"/lib64)
+export LIBRARY_PATH=\$(qp_prepend_export "LIBRARY_PATH" "\${QP_ROOT}"/lib:"\${QP_ROOT}"/extra_lib:"\${QP_ROOT}"/lib64)
+export C_INCLUDE_PATH=\$(qp_prepend_export "C_INCLUDE_PATH" "\${QP_ROOT}"/include)
+
+# export QP_NIC=ib0
 EOF
 
 #exit 0