mirror of
https://github.com/QuantumPackage/qp2.git
synced 2024-11-06 21:43:39 +01:00
Optimizations
This commit is contained in:
parent
08d197ebbb
commit
bca317a415
@ -949,6 +949,9 @@ subroutine get_d2(gen, phasemask, bannedOrb, banned, mat, mask, h, p, sp, coefs)
|
|||||||
if(tip == 3) then
|
if(tip == 3) then
|
||||||
puti = p(1, mi)
|
puti = p(1, mi)
|
||||||
if(bannedOrb(puti, mi)) return
|
if(bannedOrb(puti, mi)) return
|
||||||
|
h1 = h(1, ma)
|
||||||
|
h2 = h(2, ma)
|
||||||
|
|
||||||
do i = 1, 3
|
do i = 1, 3
|
||||||
putj = p(i, ma)
|
putj = p(i, ma)
|
||||||
if(banned(putj,puti,bant)) cycle
|
if(banned(putj,puti,bant)) cycle
|
||||||
@ -956,19 +959,21 @@ subroutine get_d2(gen, phasemask, bannedOrb, banned, mat, mask, h, p, sp, coefs)
|
|||||||
i2 = turn3(2,i)
|
i2 = turn3(2,i)
|
||||||
p1 = p(i1, ma)
|
p1 = p(i1, ma)
|
||||||
p2 = p(i2, ma)
|
p2 = p(i2, ma)
|
||||||
h1 = h(1, ma)
|
|
||||||
h2 = h(2, ma)
|
|
||||||
|
|
||||||
hij = (mo_two_e_integral(p1, p2, h1, h2) - mo_two_e_integral(p2,p1, h1, h2)) * get_phase_bi(phasemask, ma, ma, h1, p1, h2, p2, N_int)
|
hij = mo_two_e_integral(p1, p2, h1, h2) - mo_two_e_integral(p2, p1, h1, h2)
|
||||||
|
if (hij == 0.d0) cycle
|
||||||
|
|
||||||
|
hij = hij * get_phase_bi(phasemask, ma, ma, h1, p1, h2, p2, N_int)
|
||||||
|
|
||||||
if(ma == 1) then
|
if(ma == 1) then
|
||||||
!DIR$ NOVECTOR
|
!DIR$ LOOP COUNT AVG(4)
|
||||||
do k=1,N_states
|
do k=1,N_states
|
||||||
mat(k, putj, puti) = mat(k, putj, puti) +coefs(k) * hij
|
mat(k, putj, puti) = mat(k, putj, puti) + coefs(k) * hij
|
||||||
enddo
|
enddo
|
||||||
else
|
else
|
||||||
!DIR$ NOVECTOR
|
!DIR$ LOOP COUNT AVG(4)
|
||||||
do k=1,N_states
|
do k=1,N_states
|
||||||
mat(k, puti, putj) = mat(k, puti, putj) +coefs(k) * hij
|
mat(k, puti, putj) = mat(k, puti, putj) + coefs(k) * hij
|
||||||
enddo
|
enddo
|
||||||
end if
|
end if
|
||||||
end do
|
end do
|
||||||
@ -985,11 +990,14 @@ subroutine get_d2(gen, phasemask, bannedOrb, banned, mat, mask, h, p, sp, coefs)
|
|||||||
if(banned(puti,putj,bant) .or. bannedOrb(puti,1)) cycle
|
if(banned(puti,putj,bant) .or. bannedOrb(puti,1)) cycle
|
||||||
p1 = p(turn2(i), 1)
|
p1 = p(turn2(i), 1)
|
||||||
|
|
||||||
hij = mo_two_e_integral(p1, p2, h1, h2) * get_phase_bi(phasemask, 1, 2, h1, p1, h2, p2, N_int)
|
hij = mo_two_e_integral(p1, p2, h1, h2)
|
||||||
!DIR$ NOVECTOR
|
if (hij /= 0.d0) then
|
||||||
|
hij = hij * get_phase_bi(phasemask, 1, 2, h1, p1, h2, p2, N_int)
|
||||||
|
!DIR$ LOOP COUNT AVG(4)
|
||||||
do k=1,N_states
|
do k=1,N_states
|
||||||
mat(k, puti, putj) = mat(k, puti, putj) +coefs(k) * hij
|
mat(k, puti, putj) = mat(k, puti, putj) + coefs(k) * hij
|
||||||
enddo
|
enddo
|
||||||
|
endif
|
||||||
end do
|
end do
|
||||||
end do
|
end do
|
||||||
end if
|
end if
|
||||||
@ -1010,8 +1018,11 @@ subroutine get_d2(gen, phasemask, bannedOrb, banned, mat, mask, h, p, sp, coefs)
|
|||||||
i2 = turn2d(2, i, j)
|
i2 = turn2d(2, i, j)
|
||||||
p1 = p(i1, ma)
|
p1 = p(i1, ma)
|
||||||
p2 = p(i2, ma)
|
p2 = p(i2, ma)
|
||||||
hij = (mo_two_e_integral(p1, p2, h1, h2) - mo_two_e_integral(p2,p1, h1, h2)) * get_phase_bi(phasemask, ma, ma, h1, p1, h2, p2, N_int)
|
hij = mo_two_e_integral(p1, p2, h1, h2) - mo_two_e_integral(p2,p1, h1, h2)
|
||||||
!DIR$ NOVECTOR
|
if (hij == 0.d0) cycle
|
||||||
|
|
||||||
|
hij = hij * get_phase_bi(phasemask, ma, ma, h1, p1, h2, p2, N_int)
|
||||||
|
!DIR$ LOOP COUNT AVG(4)
|
||||||
do k=1,N_states
|
do k=1,N_states
|
||||||
mat(k, puti, putj) = mat(k, puti, putj) +coefs(k) * hij
|
mat(k, puti, putj) = mat(k, puti, putj) +coefs(k) * hij
|
||||||
enddo
|
enddo
|
||||||
@ -1029,14 +1040,17 @@ subroutine get_d2(gen, phasemask, bannedOrb, banned, mat, mask, h, p, sp, coefs)
|
|||||||
if(banned(puti,putj,1)) cycle
|
if(banned(puti,putj,1)) cycle
|
||||||
p2 = p(i, ma)
|
p2 = p(i, ma)
|
||||||
|
|
||||||
hij = mo_two_e_integral(p1, p2, h1, h2) * get_phase_bi(phasemask, mi, ma, h1, p1, h2, p2, N_int)
|
hij = mo_two_e_integral(p1, p2, h1, h2)
|
||||||
|
if (hij == 0.d0) cycle
|
||||||
|
|
||||||
|
hij = hij * get_phase_bi(phasemask, mi, ma, h1, p1, h2, p2, N_int)
|
||||||
if (puti < putj) then
|
if (puti < putj) then
|
||||||
!DIR$ NOVECTOR
|
!DIR$ LOOP COUNT AVG(4)
|
||||||
do k=1,N_states
|
do k=1,N_states
|
||||||
mat(k, puti, putj) = mat(k, puti, putj) + coefs(k) * hij
|
mat(k, puti, putj) = mat(k, puti, putj) + coefs(k) * hij
|
||||||
enddo
|
enddo
|
||||||
else
|
else
|
||||||
!DIR$ NOVECTOR
|
!DIR$ LOOP COUNT AVG(4)
|
||||||
do k=1,N_states
|
do k=1,N_states
|
||||||
mat(k, putj, puti) = mat(k, putj, puti) + coefs(k) * hij
|
mat(k, putj, puti) = mat(k, putj, puti) + coefs(k) * hij
|
||||||
enddo
|
enddo
|
||||||
@ -1050,14 +1064,17 @@ subroutine get_d2(gen, phasemask, bannedOrb, banned, mat, mask, h, p, sp, coefs)
|
|||||||
p2 = p(2, mi)
|
p2 = p(2, mi)
|
||||||
h1 = h(1, mi)
|
h1 = h(1, mi)
|
||||||
h2 = h(2, mi)
|
h2 = h(2, mi)
|
||||||
hij = (mo_two_e_integral(p1, p2, h1, h2) - mo_two_e_integral(p2,p1, h1, h2)) * get_phase_bi(phasemask, mi, mi, h1, p1, h2, p2, N_int)
|
hij = (mo_two_e_integral(p1, p2, h1, h2) - mo_two_e_integral(p2,p1, h1, h2))
|
||||||
!DIR$ NOVECTOR
|
if (hij /= 0.d0) then
|
||||||
|
hij = hij * get_phase_bi(phasemask, mi, mi, h1, p1, h2, p2, N_int)
|
||||||
|
!DIR$ LOOP COUNT AVG(4)
|
||||||
do k=1,N_states
|
do k=1,N_states
|
||||||
mat(k, puti, putj) = mat(k, puti, putj) +coefs(k) * hij
|
mat(k, puti, putj) = mat(k, puti, putj) + coefs(k) * hij
|
||||||
enddo
|
enddo
|
||||||
end if
|
end if
|
||||||
end if
|
end if
|
||||||
end if
|
end if
|
||||||
|
end if
|
||||||
end
|
end
|
||||||
|
|
||||||
|
|
||||||
@ -1121,7 +1138,10 @@ subroutine get_d1(gen, phasemask, bannedOrb, banned, mat, mask, h, p, sp, coefs)
|
|||||||
hij = hij_cache(putj,1) - hij_cache(putj,2)
|
hij = hij_cache(putj,1) - hij_cache(putj,2)
|
||||||
if (hij /= 0.d0) then
|
if (hij /= 0.d0) then
|
||||||
hij = hij * get_phase_bi(phasemask, ma, ma, putj, p1, hfix, p2, N_int)
|
hij = hij * get_phase_bi(phasemask, ma, ma, putj, p1, hfix, p2, N_int)
|
||||||
tmp_row(1:N_states,putj) = tmp_row(1:N_states,putj) + hij * coefs(1:N_states)
|
!DIR$ LOOP COUNT AVG(4)
|
||||||
|
do k=1,N_states
|
||||||
|
tmp_row(k,putj) = tmp_row(k,putj) + hij * coefs(k)
|
||||||
|
enddo
|
||||||
endif
|
endif
|
||||||
end do
|
end do
|
||||||
do putj=hfix+1, mo_num
|
do putj=hfix+1, mo_num
|
||||||
@ -1130,7 +1150,10 @@ subroutine get_d1(gen, phasemask, bannedOrb, banned, mat, mask, h, p, sp, coefs)
|
|||||||
hij = hij_cache(putj,2) - hij_cache(putj,1)
|
hij = hij_cache(putj,2) - hij_cache(putj,1)
|
||||||
if (hij /= 0.d0) then
|
if (hij /= 0.d0) then
|
||||||
hij = hij * get_phase_bi(phasemask, ma, ma, hfix, p1, putj, p2, N_int)
|
hij = hij * get_phase_bi(phasemask, ma, ma, hfix, p1, putj, p2, N_int)
|
||||||
tmp_row(1:N_states,putj) = tmp_row(1:N_states,putj) + hij * coefs(1:N_states)
|
!DIR$ LOOP COUNT AVG(4)
|
||||||
|
do k=1,N_states
|
||||||
|
tmp_row(k,putj) = tmp_row(k,putj) + hij * coefs(k)
|
||||||
|
enddo
|
||||||
endif
|
endif
|
||||||
end do
|
end do
|
||||||
|
|
||||||
@ -1138,7 +1161,7 @@ subroutine get_d1(gen, phasemask, bannedOrb, banned, mat, mask, h, p, sp, coefs)
|
|||||||
mat(1:N_states,1:mo_num,puti) = mat(1:N_states,1:mo_num,puti) + tmp_row(1:N_states,1:mo_num)
|
mat(1:N_states,1:mo_num,puti) = mat(1:N_states,1:mo_num,puti) + tmp_row(1:N_states,1:mo_num)
|
||||||
else
|
else
|
||||||
do l=1,mo_num
|
do l=1,mo_num
|
||||||
!DIR$ NOVECTOR
|
!DIR$ LOOP COUNT AVG(4)
|
||||||
do k=1,N_states
|
do k=1,N_states
|
||||||
mat(k,puti,l) = mat(k,puti,l) + tmp_row(k,l)
|
mat(k,puti,l) = mat(k,puti,l) + tmp_row(k,l)
|
||||||
enddo
|
enddo
|
||||||
@ -1161,7 +1184,7 @@ subroutine get_d1(gen, phasemask, bannedOrb, banned, mat, mask, h, p, sp, coefs)
|
|||||||
hij = hij_cache(puti,2)
|
hij = hij_cache(puti,2)
|
||||||
if (hij /= 0.d0) then
|
if (hij /= 0.d0) then
|
||||||
hij = hij * get_phase_bi(phasemask, ma, mi, hfix, p2, puti, pfix, N_int)
|
hij = hij * get_phase_bi(phasemask, ma, mi, hfix, p2, puti, pfix, N_int)
|
||||||
!DIR$ NOVECTOR
|
!DIR$ LOOP COUNT AVG(4)
|
||||||
do k=1,N_states
|
do k=1,N_states
|
||||||
tmp_row(k,puti) = tmp_row(k,puti) + hij * coefs(k)
|
tmp_row(k,puti) = tmp_row(k,puti) + hij * coefs(k)
|
||||||
enddo
|
enddo
|
||||||
@ -1186,8 +1209,8 @@ subroutine get_d1(gen, phasemask, bannedOrb, banned, mat, mask, h, p, sp, coefs)
|
|||||||
mat(:,:,p1) = mat(:,:,p1) + tmp_row(:,:)
|
mat(:,:,p1) = mat(:,:,p1) + tmp_row(:,:)
|
||||||
mat(:,:,p2) = mat(:,:,p2) + tmp_row2(:,:)
|
mat(:,:,p2) = mat(:,:,p2) + tmp_row2(:,:)
|
||||||
else
|
else
|
||||||
!DIR$ NOVECTOR
|
|
||||||
do l=1,mo_num
|
do l=1,mo_num
|
||||||
|
!DIR$ LOOP COUNT AVG(4)
|
||||||
do k=1,N_states
|
do k=1,N_states
|
||||||
mat(k,p1,l) = mat(k,p1,l) + tmp_row(k,l)
|
mat(k,p1,l) = mat(k,p1,l) + tmp_row(k,l)
|
||||||
mat(k,p2,l) = mat(k,p2,l) + tmp_row2(k,l)
|
mat(k,p2,l) = mat(k,p2,l) + tmp_row2(k,l)
|
||||||
@ -1227,7 +1250,7 @@ subroutine get_d1(gen, phasemask, bannedOrb, banned, mat, mask, h, p, sp, coefs)
|
|||||||
|
|
||||||
mat(:, :puti-1, puti) = mat(:, :puti-1, puti) + tmp_row(:,:puti-1)
|
mat(:, :puti-1, puti) = mat(:, :puti-1, puti) + tmp_row(:,:puti-1)
|
||||||
do l=puti,mo_num
|
do l=puti,mo_num
|
||||||
!DIR$ NOVECTOR
|
!DIR$ LOOP COUNT AVG(4)
|
||||||
do k=1,N_states
|
do k=1,N_states
|
||||||
mat(k, puti, l) = mat(k, puti,l) + tmp_row(k,l)
|
mat(k, puti, l) = mat(k, puti,l) + tmp_row(k,l)
|
||||||
enddo
|
enddo
|
||||||
@ -1250,6 +1273,7 @@ subroutine get_d1(gen, phasemask, bannedOrb, banned, mat, mask, h, p, sp, coefs)
|
|||||||
hij = hij_cache(puti,1)
|
hij = hij_cache(puti,1)
|
||||||
if (hij /= 0.d0) then
|
if (hij /= 0.d0) then
|
||||||
hij = hij * get_phase_bi(phasemask, mi, ma, hfix, pfix, puti, p1, N_int)
|
hij = hij * get_phase_bi(phasemask, mi, ma, hfix, pfix, puti, p1, N_int)
|
||||||
|
!DIR$ LOOP COUNT AVG(4)
|
||||||
do k=1,N_states
|
do k=1,N_states
|
||||||
tmp_row(k,puti) = tmp_row(k,puti) + hij * coefs(k)
|
tmp_row(k,puti) = tmp_row(k,puti) + hij * coefs(k)
|
||||||
enddo
|
enddo
|
||||||
@ -1269,14 +1293,14 @@ subroutine get_d1(gen, phasemask, bannedOrb, banned, mat, mask, h, p, sp, coefs)
|
|||||||
end do
|
end do
|
||||||
mat(:,:p2-1,p2) = mat(:,:p2-1,p2) + tmp_row(:,:p2-1)
|
mat(:,:p2-1,p2) = mat(:,:p2-1,p2) + tmp_row(:,:p2-1)
|
||||||
do l=p2,mo_num
|
do l=p2,mo_num
|
||||||
!DIR$ NOVECTOR
|
!DIR$ LOOP COUNT AVG(4)
|
||||||
do k=1,N_states
|
do k=1,N_states
|
||||||
mat(k,p2,l) = mat(k,p2,l) + tmp_row(k,l)
|
mat(k,p2,l) = mat(k,p2,l) + tmp_row(k,l)
|
||||||
enddo
|
enddo
|
||||||
enddo
|
enddo
|
||||||
mat(:,:p1-1,p1) = mat(:,:p1-1,p1) + tmp_row2(:,:p1-1)
|
mat(:,:p1-1,p1) = mat(:,:p1-1,p1) + tmp_row2(:,:p1-1)
|
||||||
do l=p1,mo_num
|
do l=p1,mo_num
|
||||||
!DIR$ NOVECTOR
|
!DIR$ LOOP COUNT AVG(4)
|
||||||
do k=1,N_states
|
do k=1,N_states
|
||||||
mat(k,p1,l) = mat(k,p1,l) + tmp_row2(k,l)
|
mat(k,p1,l) = mat(k,p1,l) + tmp_row2(k,l)
|
||||||
enddo
|
enddo
|
||||||
@ -1303,7 +1327,7 @@ subroutine get_d1(gen, phasemask, bannedOrb, banned, mat, mask, h, p, sp, coefs)
|
|||||||
if(bannedOrb(p1, s1) .or. bannedOrb(p2, s2) .or. banned(p1, p2, 1)) cycle
|
if(bannedOrb(p1, s1) .or. bannedOrb(p2, s2) .or. banned(p1, p2, 1)) cycle
|
||||||
call apply_particles(mask, s1, p1, s2, p2, det, ok, N_int)
|
call apply_particles(mask, s1, p1, s2, p2, det, ok, N_int)
|
||||||
call i_h_j(gen, det, N_int, hij)
|
call i_h_j(gen, det, N_int, hij)
|
||||||
!DIR$ NOVECTOR
|
!DIR$ LOOP COUNT AVG(4)
|
||||||
do k=1,N_states
|
do k=1,N_states
|
||||||
mat(k, p1, p2) = mat(k, p1, p2) + coefs(k) * hij
|
mat(k, p1, p2) = mat(k, p1, p2) + coefs(k) * hij
|
||||||
enddo
|
enddo
|
||||||
@ -1350,35 +1374,15 @@ subroutine get_d0(gen, phasemask, bannedOrb, banned, mat, mask, h, p, sp, coefs)
|
|||||||
call i_h_j(gen, det, N_int, hij)
|
call i_h_j(gen, det, N_int, hij)
|
||||||
else
|
else
|
||||||
phase = get_phase_bi(phasemask, 1, 2, h1, p1, h2, p2, N_int)
|
phase = get_phase_bi(phasemask, 1, 2, h1, p1, h2, p2, N_int)
|
||||||
! hij = mo_two_e_integral(p2, p1, h2, h1) * phase
|
|
||||||
hij = hij_cache1(p2) * phase
|
hij = hij_cache1(p2) * phase
|
||||||
end if
|
end if
|
||||||
if (hij == 0.d0) cycle
|
if (hij == 0.d0) cycle
|
||||||
!DIR$ NOVECTOR
|
!DIR$ LOOP COUNT AVG(4)
|
||||||
do k=1,N_states
|
do k=1,N_states
|
||||||
mat(k, p1, p2) = mat(k, p1, p2) + coefs(k) * hij ! HOTSPOT
|
mat(k, p1, p2) = mat(k, p1, p2) + coefs(k) * hij ! HOTSPOT
|
||||||
enddo
|
enddo
|
||||||
end do
|
end do
|
||||||
end do
|
end do
|
||||||
! do p2=1, mo_num
|
|
||||||
! if(bannedOrb(p2,2)) cycle
|
|
||||||
! call get_mo_two_e_integrals(p2,h1,h2,mo_num,hij_cache1,mo_integrals_map)
|
|
||||||
! do p1=1, mo_num
|
|
||||||
! if(bannedOrb(p1, 1) .or. banned(p1, p2, bant)) cycle
|
|
||||||
! if(p1 /= h1 .and. p2 /= h2) then
|
|
||||||
! if (hij_cache1(p1) == 0.d0) cycle
|
|
||||||
! phase = get_phase_bi(phasemask, 1, 2, h1, p1, h2, p2, N_int)
|
|
||||||
! hij = hij_cache1(p1) * phase
|
|
||||||
! else
|
|
||||||
! call apply_particles(mask, 1,p1,2,p2, det, ok, N_int)
|
|
||||||
! call i_h_j(gen, det, N_int, hij)
|
|
||||||
! if (hij == 0.d0) cycle
|
|
||||||
! end if
|
|
||||||
! do k=1,N_states
|
|
||||||
! mat(k, p1, p2) = mat(k, p1, p2) + coefs(k) * hij ! HOTSPOT
|
|
||||||
! enddo
|
|
||||||
! end do
|
|
||||||
! end do
|
|
||||||
|
|
||||||
else ! AA BB
|
else ! AA BB
|
||||||
p1 = p(1,sp)
|
p1 = p(1,sp)
|
||||||
@ -1393,32 +1397,16 @@ subroutine get_d0(gen, phasemask, bannedOrb, banned, mat, mask, h, p, sp, coefs)
|
|||||||
if(puti == p1 .or. putj == p2 .or. puti == p2 .or. putj == p1) then
|
if(puti == p1 .or. putj == p2 .or. puti == p2 .or. putj == p1) then
|
||||||
call apply_particles(mask, sp,puti,sp,putj, det, ok, N_int)
|
call apply_particles(mask, sp,puti,sp,putj, det, ok, N_int)
|
||||||
call i_h_j(gen, det, N_int, hij)
|
call i_h_j(gen, det, N_int, hij)
|
||||||
else
|
|
||||||
hij = (mo_two_e_integral(p1, p2, puti, putj) - mo_two_e_integral(p2, p1, puti, putj))* get_phase_bi(phasemask, sp, sp, puti, p1 , putj, p2, N_int)
|
|
||||||
end if
|
|
||||||
if (hij == 0.d0) cycle
|
if (hij == 0.d0) cycle
|
||||||
!DIR$ NOVECTOR
|
else
|
||||||
|
hij = (mo_two_e_integral(p1, p2, puti, putj) - mo_two_e_integral(p2, p1, puti, putj))
|
||||||
|
if (hij == 0.d0) cycle
|
||||||
|
hij = hij * get_phase_bi(phasemask, sp, sp, puti, p1 , putj, p2, N_int)
|
||||||
|
end if
|
||||||
|
!DIR$ LOOP COUNT AVG(4)
|
||||||
do k=1,N_states
|
do k=1,N_states
|
||||||
mat(k, puti, putj) = mat(k, puti, putj) + coefs(k) * hij
|
mat(k, puti, putj) = mat(k, puti, putj) + coefs(k) * hij
|
||||||
enddo
|
enddo
|
||||||
! if(bannedOrb(putj, sp) .or. banned(putj, sp, bant)) cycle
|
|
||||||
! if(puti /= p1 .and. putj /= p2 .and. puti /= p2 .and. putj /= p1) then
|
|
||||||
! hij = hij_cache1(putj) - hij_cache2(putj)
|
|
||||||
! if (hij /= 0.d0) then
|
|
||||||
! hij = hij * get_phase_bi(phasemask, sp, sp, puti, p1 , putj, p2, N_int)
|
|
||||||
! do k=1,N_states
|
|
||||||
! mat(k, puti, putj) = mat(k, puti, putj) + coefs(k) * hij
|
|
||||||
! enddo
|
|
||||||
! endif
|
|
||||||
! else
|
|
||||||
! call apply_particles(mask, sp,puti,sp,putj, det, ok, N_int)
|
|
||||||
! call i_h_j(gen, det, N_int, hij)
|
|
||||||
! if (hij /= 0.d0) then
|
|
||||||
! do k=1,N_states
|
|
||||||
! mat(k, puti, putj) = mat(k, puti, putj) + coefs(k) * hij
|
|
||||||
! enddo
|
|
||||||
! endif
|
|
||||||
! end if
|
|
||||||
end do
|
end do
|
||||||
end do
|
end do
|
||||||
end if
|
end if
|
||||||
@ -1452,8 +1440,8 @@ subroutine past_d2(banned, p, sp)
|
|||||||
integer :: i,j
|
integer :: i,j
|
||||||
|
|
||||||
if(sp == 3) then
|
if(sp == 3) then
|
||||||
do i=1,p(0,1)
|
|
||||||
do j=1,p(0,2)
|
do j=1,p(0,2)
|
||||||
|
do i=1,p(0,1)
|
||||||
banned(p(i,1), p(j,2)) = .true.
|
banned(p(i,1), p(j,2)) = .true.
|
||||||
end do
|
end do
|
||||||
end do
|
end do
|
||||||
|
@ -210,6 +210,7 @@ subroutine H_S2_u_0_nstates_openmp_work_$N_int(v_t,s_t,u_t,N_st,sze,istart,iend,
|
|||||||
integer*8 :: last_found, left, right, right_max
|
integer*8 :: last_found, left, right, right_max
|
||||||
double precision :: rss, mem, ratio
|
double precision :: rss, mem, ratio
|
||||||
double precision, allocatable :: utl(:,:)
|
double precision, allocatable :: utl(:,:)
|
||||||
|
integer, parameter :: block_size=128
|
||||||
|
|
||||||
! call resident_memory(rss)
|
! call resident_memory(rss)
|
||||||
! mem = dble(singles_beta_csc_size) / 1024.d0**3
|
! mem = dble(singles_beta_csc_size) / 1024.d0**3
|
||||||
@ -261,7 +262,7 @@ compute_singles=.True.
|
|||||||
singles_a(maxab), &
|
singles_a(maxab), &
|
||||||
singles_b(maxab), &
|
singles_b(maxab), &
|
||||||
doubles(maxab), &
|
doubles(maxab), &
|
||||||
idx(maxab), utl(N_st,32))
|
idx(maxab), utl(N_st,block_size))
|
||||||
|
|
||||||
kcol_prev=-1
|
kcol_prev=-1
|
||||||
|
|
||||||
@ -399,9 +400,9 @@ compute_singles=.True.
|
|||||||
! -----------------------
|
! -----------------------
|
||||||
|
|
||||||
!DIR$ LOOP COUNT avg(1000)
|
!DIR$ LOOP COUNT avg(1000)
|
||||||
do k = 1,n_singles_a,32
|
do k = 1,n_singles_a,block_size
|
||||||
! Prefetch u_t(:,l_a)
|
! Prefetch u_t(:,l_a)
|
||||||
do kk=0,31
|
do kk=0,block_size-1
|
||||||
if (k+kk > n_singles_a) exit
|
if (k+kk > n_singles_a) exit
|
||||||
l_a = singles_a(k+kk)
|
l_a = singles_a(k+kk)
|
||||||
ASSERT (l_a <= N_det)
|
ASSERT (l_a <= N_det)
|
||||||
@ -411,7 +412,7 @@ compute_singles=.True.
|
|||||||
enddo
|
enddo
|
||||||
enddo
|
enddo
|
||||||
|
|
||||||
do kk=0,31
|
do kk=0,block_size-1
|
||||||
if (k+kk > n_singles_a) exit
|
if (k+kk > n_singles_a) exit
|
||||||
l_a = singles_a(k+kk)
|
l_a = singles_a(k+kk)
|
||||||
lrow = psi_bilinear_matrix_rows(l_a)
|
lrow = psi_bilinear_matrix_rows(l_a)
|
||||||
@ -488,9 +489,9 @@ compute_singles=.True.
|
|||||||
|
|
||||||
tmp_det2(1:$N_int,2) = psi_det_beta_unique (1:$N_int, kcol)
|
tmp_det2(1:$N_int,2) = psi_det_beta_unique (1:$N_int, kcol)
|
||||||
!DIR$ LOOP COUNT avg(1000)
|
!DIR$ LOOP COUNT avg(1000)
|
||||||
do i=1,n_singles_a,32
|
do i=1,n_singles_a,block_size
|
||||||
! Prefetch u_t(:,l_a)
|
! Prefetch u_t(:,l_a)
|
||||||
do kk=0,31
|
do kk=0,block_size-1
|
||||||
if (i+kk > n_singles_a) exit
|
if (i+kk > n_singles_a) exit
|
||||||
l_a = singles_a(i+kk)
|
l_a = singles_a(i+kk)
|
||||||
ASSERT (l_a <= N_det)
|
ASSERT (l_a <= N_det)
|
||||||
@ -500,7 +501,7 @@ compute_singles=.True.
|
|||||||
enddo
|
enddo
|
||||||
enddo
|
enddo
|
||||||
|
|
||||||
do kk=0,31
|
do kk=0,block_size-1
|
||||||
if (i+kk > n_singles_a) exit
|
if (i+kk > n_singles_a) exit
|
||||||
l_a = singles_a(i+kk)
|
l_a = singles_a(i+kk)
|
||||||
lrow = psi_bilinear_matrix_rows(l_a)
|
lrow = psi_bilinear_matrix_rows(l_a)
|
||||||
@ -522,9 +523,9 @@ compute_singles=.True.
|
|||||||
! ----------------------------------
|
! ----------------------------------
|
||||||
|
|
||||||
!DIR$ LOOP COUNT avg(50000)
|
!DIR$ LOOP COUNT avg(50000)
|
||||||
do i=1,n_doubles,32
|
do i=1,n_doubles,block_size
|
||||||
! Prefetch u_t(:,l_a)
|
! Prefetch u_t(:,l_a)
|
||||||
do kk=0,31
|
do kk=0,block_size-1
|
||||||
if (i+kk > n_doubles) exit
|
if (i+kk > n_doubles) exit
|
||||||
l_a = doubles(i+kk)
|
l_a = doubles(i+kk)
|
||||||
ASSERT (l_a <= N_det)
|
ASSERT (l_a <= N_det)
|
||||||
@ -534,7 +535,7 @@ compute_singles=.True.
|
|||||||
enddo
|
enddo
|
||||||
enddo
|
enddo
|
||||||
|
|
||||||
do kk=0,31
|
do kk=0,block_size-1
|
||||||
if (i+kk > n_doubles) exit
|
if (i+kk > n_doubles) exit
|
||||||
l_a = doubles(i+kk)
|
l_a = doubles(i+kk)
|
||||||
lrow = psi_bilinear_matrix_rows(l_a)
|
lrow = psi_bilinear_matrix_rows(l_a)
|
||||||
@ -597,8 +598,8 @@ compute_singles=.True.
|
|||||||
|
|
||||||
tmp_det2(1:$N_int,1) = psi_det_alpha_unique(1:$N_int, krow)
|
tmp_det2(1:$N_int,1) = psi_det_alpha_unique(1:$N_int, krow)
|
||||||
!DIR$ LOOP COUNT avg(1000)
|
!DIR$ LOOP COUNT avg(1000)
|
||||||
do i=1,n_singles_b,32
|
do i=1,n_singles_b,block_size
|
||||||
do kk=0,31
|
do kk=0,block_size-1
|
||||||
if (i+kk > n_singles_b) exit
|
if (i+kk > n_singles_b) exit
|
||||||
l_b = singles_b(i+kk)
|
l_b = singles_b(i+kk)
|
||||||
ASSERT (l_b <= N_det)
|
ASSERT (l_b <= N_det)
|
||||||
@ -611,7 +612,7 @@ compute_singles=.True.
|
|||||||
enddo
|
enddo
|
||||||
enddo
|
enddo
|
||||||
|
|
||||||
do kk=0,31
|
do kk=0,block_size-1
|
||||||
if (i+kk > n_singles_b) exit
|
if (i+kk > n_singles_b) exit
|
||||||
l_b = singles_b(i+kk)
|
l_b = singles_b(i+kk)
|
||||||
l_a = psi_bilinear_matrix_transp_order(l_b)
|
l_a = psi_bilinear_matrix_transp_order(l_b)
|
||||||
@ -632,8 +633,8 @@ compute_singles=.True.
|
|||||||
! ----------------------------------
|
! ----------------------------------
|
||||||
|
|
||||||
!DIR$ LOOP COUNT avg(50000)
|
!DIR$ LOOP COUNT avg(50000)
|
||||||
do i=1,n_doubles,32
|
do i=1,n_doubles,block_size
|
||||||
do kk=0,31
|
do kk=0,block_size-1
|
||||||
if (i+kk > n_doubles) exit
|
if (i+kk > n_doubles) exit
|
||||||
l_b = doubles(i+kk)
|
l_b = doubles(i+kk)
|
||||||
ASSERT (l_b <= N_det)
|
ASSERT (l_b <= N_det)
|
||||||
@ -645,7 +646,7 @@ compute_singles=.True.
|
|||||||
enddo
|
enddo
|
||||||
enddo
|
enddo
|
||||||
|
|
||||||
do kk=0,31
|
do kk=0,block_size-1
|
||||||
if (i+kk > n_doubles) exit
|
if (i+kk > n_doubles) exit
|
||||||
l_b = doubles(i+kk)
|
l_b = doubles(i+kk)
|
||||||
l_a = psi_bilinear_matrix_transp_order(l_b)
|
l_a = psi_bilinear_matrix_transp_order(l_b)
|
||||||
|
Loading…
Reference in New Issue
Block a user