comp-chem/cp2k
1
0
Fork 0
mirror of https://github.com/cp2k/cp2k.git synced 2026-07-28 06:05:29 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

Adjusted CRITICIAL section names for Intel Fortran Compiler

Browse source 
- Revised some workarounds for Intel Compiler.
- Improved applying workaround (if needed).
This commit is contained in:
Hans Pabst 2025-11-20 12:44:23 +01:00
parent b2e41f9e62
commit b0ff3e3a03
8 changed files with 27 additions and 31 deletions
Download patch file Download diff file Expand all files Collapse all files

4
src/common/gamma.F
View file

@ -545,7 +545,7 @@ CONTAINS
"incomplete Gamma function is invalid")
END IF
!$OMP CRITICAL(init_md_ftable)
!$OMP CRITICAL(omp_init_md_ftable)
! *** Check, if the current initialization is sufficient ***
IF (nmax > current_nmax) THEN
CALL deallocate_md_ftable()
@ -553,7 +553,7 @@ CONTAINS
! *** for the Taylor series expansion ***
CALL create_md_ftable(nmax, 0.0_dp, 12.0_dp, 0.1_dp)
END IF
!$OMP END CRITICAL(init_md_ftable)
!$OMP END CRITICAL(omp_init_md_ftable)
END SUBROUTINE init_md_ftable

4
src/dbt/dbt_block.F
View file

@ -371,11 +371,11 @@ CONTAINS
IF (checker_tr(ind_2d(1), ind_2d(2))) CYCLE
IF (ind_2d(1) > ind_2d(2)) CALL swap(ind_2d(1), ind_2d(2))
END IF
!$OMP CRITICAL(dbt_reserve_blocks_tensor_to_matrix)
!$OMP CRITICAL(omp_dbt_reserve_blocks_tensor_to_matrix)
iblk = iblk + 1
blk_ind_1(iblk) = ind_2d(1)
blk_ind_2(iblk) = ind_2d(2)
!$OMP END CRITICAL(dbt_reserve_blocks_tensor_to_matrix)
!$OMP END CRITICAL(omp_dbt_reserve_blocks_tensor_to_matrix)
END DO
CALL dbt_iterator_stop(iter)
!$OMP END PARALLEL

4
src/dbt/dbt_reshape_ops.F
View file

@ -116,12 +116,12 @@ CONTAINS
ndata_send_mythread(iproc) = ndata_send_mythread(iproc) + PRODUCT(blk_size)
END DO
CALL dbt_iterator_stop(iter)
!$OMP CRITICAL(dbt_reshape)
!$OMP CRITICAL(omp_dbt_reshape)
nblks_send_total(:) = nblks_send_total(:) + nblks_send_mythread(:)
ndata_send_total(:) = ndata_send_total(:) + ndata_send_mythread(:)
nblks_send_mythread(:) = nblks_send_total(:) ! current totals indicate slot for this thread
ndata_send_mythread(:) = ndata_send_total(:)
!$OMP END CRITICAL(dbt_reshape)
!$OMP END CRITICAL(omp_dbt_reshape)
!$OMP BARRIER
!$OMP MASTER

16
src/eip_silicon.F
View file

@ -1184,10 +1184,10 @@ CONTAINS
! PARALLEL CASE
! create temporary private scalars for reduction sum on energies and
! temporary private array for reduction sum on forces
!$OMP CRITICAL(eip_bazant_silicon)
!$OMP CRITICAL(omp_eip_bazant_silicon)
ALLOCATE (txyz(3, nat), s2(max_nbrs, 8), s3(max_nbrs, 7), sz(max_nbrs, 6), &
num2(max_nbrs), num3(max_nbrs), numz(max_nbrs))
!$OMP END CRITICAL(eip_bazant_silicon)
!$OMP END CRITICAL(omp_eip_bazant_silicon)
IF (iam == 0) THEN
ener = 0.e0_dp
ener2 = 0.e0_dp
@ -1213,7 +1213,7 @@ CONTAINS
num2, s3(1, 1), s3(1, 2), s3(1, 3), s3(1, 4), s3(1, 5), s3(1, 6), s3(1, 7), &
num3, sz(1, 1), sz(1, 2), sz(1, 3), sz(1, 4), sz(1, 5), sz(1, 6), numz)
!$OMP CRITICAL(eip_bazant_silicon)
!$OMP CRITICAL(omp_eip_bazant_silicon)
ener = ener + tener
ener2 = ener2 + tener2
coord = coord + tcoord
@ -1225,7 +1225,7 @@ CONTAINS
fxyz(3, iat) = fxyz(3, iat) + txyz(3, iat)
END DO
DEALLOCATE (txyz, s2, s3, sz, num2, num3, numz)
!$OMP END CRITICAL(eip_bazant_silicon)
!$OMP END CRITICAL(omp_eip_bazant_silicon)
ELSE
! SERIAL CASE
@ -2349,9 +2349,9 @@ CONTAINS
! PARALLEL CASE
! create temporary private scalars for reduction sum on energies and
! temporary private array for reduction sum on forces
!$OMP CRITICAL(eip_lenosky_silicon)
!$OMP CRITICAL(omp_eip_lenosky_silicon)
ALLOCATE (txyz(3, nat), f2ij(3, npjx), f3ij(3, npjkx), f3ik(3, npjkx))
!$OMP END CRITICAL(eip_lenosky_silicon)
!$OMP END CRITICAL(omp_eip_lenosky_silicon)
IF (iam == 0) THEN
ener = 0.e0_dp
ener2 = 0.e0_dp
@ -2373,7 +2373,7 @@ CONTAINS
! write(*,*) 'subfeniat:iat1,iat2,iam',iat1,iat2,iam
CALL subfeniat_l(iat1, iat2, nat, lsta, lstb, rel, tener, tener2, &
tcoord, tcoord2, nnbrx, txyz, f2ij, npjx, f3ij, npjkx, f3ik, istop)
!$OMP CRITICAL(eip_lenosky_silicon)
!$OMP CRITICAL(omp_eip_lenosky_silicon)
ener = ener + tener
ener2 = ener2 + tener2
coord = coord + tcoord
@ -2385,7 +2385,7 @@ CONTAINS
fxyz(3, iat) = fxyz(3, iat) + txyz(3, iat)
END DO
DEALLOCATE (txyz, f2ij, f3ij, f3ik)
!$OMP END CRITICAL(eip_lenosky_silicon)
!$OMP END CRITICAL(omp_eip_lenosky_silicon)
ELSE
! SERIAL CASE

12
src/hfx_ri.F
View file

@ -4062,11 +4062,11 @@ CONTAINS
CALL dbt_iterator_next_block(iter, ind)
CALL dbt_get_block(rho_ao_t, ind, blk, found)
IF (found) THEN
!$OMP CRITICAL(get_RI_density_coeffs)
!$OMP CRITICAL(omp_get_RI_density_coeffs)
nblks = nblks + 1
idx1(nblks) = ind(1)
idx2(nblks) = ind(2)
!$OMP END CRITICAL(get_RI_density_coeffs)
!$OMP END CRITICAL(omp_get_RI_density_coeffs)
DEALLOCATE (blk)
END IF
END DO
@ -4090,9 +4090,9 @@ CONTAINS
IF (found) THEN
ALLOCATE (blk_3d(SIZE(blk, 1), SIZE(blk, 2), 1))
blk_3d(:, :, 1) = blk(:, :)
!$OMP CRITICAL(get_RI_density_coeffs)
!$OMP CRITICAL(omp_get_RI_density_coeffs)
CALL dbt_put_block(rho_ao_t_3d, [ind(1), ind(2), 1], [SIZE(blk, 1), SIZE(blk, 2), 1], blk_3d)
!$OMP END CRITICAL(get_RI_density_coeffs)
!$OMP END CRITICAL(omp_get_RI_density_coeffs)
DEALLOCATE (blk, blk_3d)
END IF
END DO
@ -4202,9 +4202,9 @@ CONTAINS
IF (found) THEN
idx = SUM(ri_data%bsizes_RI_split(1:ind(1) - 1))
!$OMP CRITICAL(get_RI_density_coeffs)
!$OMP CRITICAL(omp_get_RI_density_coeffs)
density_coeffs(idx + 1:idx + ri_data%bsizes_RI_split(ind(1))) = blk(:, 1)
!$OMP END CRITICAL(get_RI_density_coeffs)
!$OMP END CRITICAL(omp_get_RI_density_coeffs)
DEALLOCATE (blk)
END IF
END DO

4
src/qs_o3c_methods.F
View file

@ -463,10 +463,10 @@ CONTAINS
! to access the dbcsr block at the same time. Prevent that by CRITICAL section but keep
! computations before hand in order to retain speed
!$OMP CRITICAL(contract3_o3c)
!$OMP CRITICAL(omp_contract3_o3c)
CALL dbcsr_get_block_p(matrix=matrix, row=irow, col=icol, BLOCK=pblock, found=found)
CALL daxpy(s1*s2, 1.0_dp, work(:, :), 1, pblock(:, :), 1)
!$OMP END CRITICAL(contract3_o3c)
!$OMP END CRITICAL(omp_contract3_o3c)
DEALLOCATE (work)
END IF

4
src/xas_tdp_integrals.F
View file

@ -606,9 +606,9 @@ CONTAINS
END DO !iset
!Add the integral to the proper tensor block
!$OMP CRITICAL
!$OMP CRITICAL(omp_fill_pqX_tensor)
CALL dbt_put_block(pq_X, [iatom, jatom, katom], SHAPE(iabc), iabc, summation=.TRUE.)
!$OMP END CRITICAL
!$OMP END CRITICAL(omp_fill_pqX_tensor)
DEALLOCATE (iabc)
END DO !o3c_iterator

10
tools/toolchain/scripts/generate_arch_files.sh
View file

@ -400,14 +400,10 @@ EOF
cat << EOF >> $__filename
#
# Required due to memory leak that occurs if high optimisations are used
mp2_optimize_ri_basis.o: mp2_optimize_ri_basis.F
\\\$(FC) -c \\\$(subst -O2,-O0,\\\$(FCFLAGS)) \\\$<
#mp2_optimize_ri_basis.o: FCFLAGS += -O0
# Required due to SEGFAULTS occurring for higher optimisation levels
paw_basis_types.o: paw_basis_types.F
\\\$(FC) -c \\\$(subst -O2,-O1,\\\$(FCFLAGS)) \\\$<
# Reduce compilation time
hfx_contraction_methods.o: hfx_contraction_methods.F
\\\$(FC) -c \\\$(subst -O2,-O1,\\\$(FCFLAGS)) \\\$<
hfx_contraction_methods.o: FCFLAGS += -O1
#paw_basis_types.o: FCFLAGS += -O1
EOF
fi
# replace variable values in output file using eval