LCPQ/quantum_package
10
0
Fork 0
mirror of https://github.com/LCPQ/quantum_package synced 2024-06-02 11:25:26 +02:00
Code Issues Releases Wiki Activity

Acceleration : parallelism of PT2

Browse Source
This commit is contained in:
Anthony Scemama 2014-09-05 23:55:03 +02:00
parent 563a8bea4d
commit b2e65031e6
2 changed files with 15 additions and 21 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
scripts/generate_h_apply.py
View File

@ -45,7 +45,7 @@ class H_apply(object):
!$OMP accu,i_a,hole_tmp,particle_tmp,occ_particle_tmp, &
!$OMP occ_hole_tmp,key_idx,i_b,j_b,key,N_elec_in_key_part_1,&
!$OMP N_elec_in_key_hole_1,N_elec_in_key_part_2, &
!$OMP N_elec_in_key_hole_2,ia_ja_pairs,iproc) &
!$OMP N_elec_in_key_hole_2,ia_ja_pairs) &
!$OMP SHARED(key_in,N_int,elec_num_tab,mo_tot_num, &
!$OMP hole_1, particl_1, hole_2, particl_2, &
!$OMP elec_alpha_num,i_generator)"""

34
src/Dets/H_apply_template.f
View File

@ -1,4 +1,4 @@
subroutine $subroutine_diexc(key_in, hole_1,particl_1, hole_2, particl_2, i_generator $parameters )
subroutine $subroutine_diexc(key_in, hole_1,particl_1, hole_2, particl_2, i_generator, iproc $parameters )
use omp_lib
use bitmasks
implicit none
@ -14,6 +14,7 @@ subroutine $subroutine_diexc(key_in, hole_1,particl_1, hole_2, particl_2, i_gene
integer(bit_kind),allocatable :: keys_out(:,:,:)
integer(bit_kind), intent(in) :: hole_1(N_int,2), particl_1(N_int,2)
integer(bit_kind), intent(in) :: hole_2(N_int,2), particl_2(N_int,2)
integer, intent(in) :: iproc
integer(bit_kind), allocatable :: hole_save(:,:)
integer(bit_kind), allocatable :: key(:,:),hole(:,:), particle(:,:)
integer(bit_kind), allocatable :: hole_tmp(:,:), particle_tmp(:,:)
@ -28,7 +29,6 @@ subroutine $subroutine_diexc(key_in, hole_1,particl_1, hole_2, particl_2, i_gene
integer, allocatable :: ia_ja_pairs(:,:,:)
integer, allocatable :: ib_jb_pairs(:,:)
double precision :: diag_H_mat_elem
integer :: iproc
integer(omp_lock_kind), save :: lck, ifirst=0
if (ifirst == 0) then
!$ call omp_init_lock(lck)
@ -37,9 +37,7 @@ subroutine $subroutine_diexc(key_in, hole_1,particl_1, hole_2, particl_2, i_gene
$initialization
iproc = 0
$omp_parallel
!$ iproc = omp_get_thread_num()
allocate (keys_out(N_int,2,size_max), hole_save(N_int,2), &
key(N_int,2),hole(N_int,2), particle(N_int,2), hole_tmp(N_int,2),&
particle_tmp(N_int,2), occ_particle(N_int*bit_kind_size,2), &
@ -242,7 +240,7 @@ subroutine $subroutine_diexc(key_in, hole_1,particl_1, hole_2, particl_2, i_gene
$finalization
end
subroutine $subroutine_monoexc(key_in, hole_1,particl_1,i_generator $parameters )
subroutine $subroutine_monoexc(key_in, hole_1,particl_1,i_generator,iproc $parameters )
use omp_lib
use bitmasks
implicit none
@ -256,6 +254,7 @@ subroutine $subroutine_monoexc(key_in, hole_1,particl_1,i_generator $parameters
integer ,intent(in) :: i_generator
integer(bit_kind),intent(in) :: key_in(N_int,2)
integer(bit_kind),intent(in) :: hole_1(N_int,2), particl_1(N_int,2)
integer, intent(in) :: iproc
integer(bit_kind),allocatable :: keys_out(:,:,:)
integer(bit_kind),allocatable :: hole_save(:,:)
integer(bit_kind),allocatable :: key(:,:),hole(:,:), particle(:,:)
@ -272,7 +271,6 @@ subroutine $subroutine_monoexc(key_in, hole_1,particl_1,i_generator $parameters
integer, allocatable :: ia_ja_pairs(:,:,:)
logical, allocatable :: array_pairs(:,:)
double precision :: diag_H_mat_elem
integer :: iproc
integer(omp_lock_kind), save :: lck, ifirst=0
if (ifirst == 0) then
ifirst=1
@ -281,9 +279,7 @@ subroutine $subroutine_monoexc(key_in, hole_1,particl_1,i_generator $parameters
$initialization
iproc = 0
$omp_parallel
!$ iproc = omp_get_thread_num()
allocate (keys_out(N_int,2,size_max), hole_save(N_int,2), &
key(N_int,2),hole(N_int,2), particle(N_int,2), hole_tmp(N_int,2),&
particle_tmp(N_int,2), occ_particle(N_int*bit_kind_size,2), &
@ -379,6 +375,7 @@ subroutine $subroutine($params_main)
integer(omp_lock_kind) :: lck
integer(bit_kind), allocatable :: mask(:,:,:)
integer :: ispin, k
integer :: iproc
PROVIDE H_apply_buffer_allocated mo_bielec_integrals_in_map N_det_selectors psi_generators
PROVIDE psi_det_sorted_bit coef_hf_selector psi_det psi_coef ref_bitmask_energy
@ -389,9 +386,10 @@ subroutine $subroutine($params_main)
!$ call omp_init_lock(lck)
IRP_IF I_LIKE_BUGS
!$OMP PARALLEL DEFAULT(SHARED) &
!$OMP PRIVATE(i_generator,wall_2,ispin,k,mask)
!$OMP PRIVATE(i_generator,wall_2,ispin,k,mask,iproc)
iproc = 0
!$ iproc = omp_get_thread_num()
allocate( mask(N_int,2,6) )
!$OMP DO SCHEDULE(dynamic,4)
do i_generator=1,nmax
@ -428,12 +426,12 @@ IRP_IF I_LIKE_BUGS
call $subroutine_diexc(psi_generators(1,1,i_generator), &
mask(1,1,d_hole1), mask(1,1,d_part1), &
mask(1,1,d_hole2), mask(1,1,d_part2), &
i_generator $params_post)
i_generator, iproc $params_post)
endif
if($do_mono_excitations)then
call $subroutine_monoexc(psi_generators(1,1,i_generator), &
mask(1,1,s_hole ), mask(1,1,s_part ), &
i_generator $params_post)
i_generator, iproc $params_post)
endif
!$ call omp_set_lock(lck)
call wall_time(wall_2)
@ -449,14 +447,10 @@ IRP_IF I_LIKE_BUGS
!$OMP END PARALLEL
!$ call omp_destroy_lock(lck)
allocate( mask(N_int,2,6) )
! do i_generator=1,N_det_generators
do i_generator=nmax+1,N_det_generators
IRP_ELSE
allocate( mask(N_int,2,6) )
do i_generator=1,N_det_generators
IRP_ENDIF
if (abort_here) then
exit
@ -490,12 +484,12 @@ IRP_ENDIF
call $subroutine_diexc(psi_generators(1,1,i_generator), &
mask(1,1,d_hole1), mask(1,1,d_part1), &
mask(1,1,d_hole2), mask(1,1,d_part2), &
i_generator $params_post)
i_generator, 0 $params_post)
endif
if($do_mono_excitations)then
call $subroutine_monoexc(psi_generators(1,1,i_generator), &
mask(1,1,s_hole ), mask(1,1,s_part ), &
i_generator $params_post)
i_generator, 0 $params_post)
endif
call wall_time(wall_2)
$printout_always