Skip to content

Fix 1D multi-rank MPI_GATHERV bug in post-process silo output #1138

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
sbryngelson merged 2 commits into from
Feb 13, 2026
+42 −19
Merged

Fix 1D multi-rank MPI_GATHERV bug in post-process silo output #1138

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
b3d4ec5
Update m_data_output.fpp
sbryngelson Feb 13, 2026
0a79011
Fix 1D multi-rank MPI_GATHERV undefined behavior in post-process
sbryngelson Feb 13, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion src/post_process/m_data_output.fpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,7 +11,7 @@ module m_data_output

use m_derived_types ! Definitions of the derived types

use m_global_parameters ! Global parameters for the code
use m_global_parameters ! Global parameters

use m_derived_variables !< Procedures used to compute quantities derived

Expand Down
59 changes: 41 additions & 18 deletions src/post_process/m_mpi_proxy.fpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -168,6 +168,7 @@ contains

#ifdef MFC_MPI
integer :: ierr !< Generic flag used to identify and report MPI errors
real(wp) :: ext_temp(0:num_procs - 1)

! Simulation is 3D
if (p > 0) then
Expand Down Expand Up @@ -273,17 +274,20 @@ contains
! Simulation is 1D
else

! For 1D, recvcounts/displs are sized for grid defragmentation
! (m+1 per rank), not for scalar gathers. Use MPI_GATHER instead.

! Minimum spatial extent in the x-direction
call MPI_GATHERV(minval(x_cb), 1, mpi_p, &
spatial_extents(1, 0), recvcounts, 4*displs, &
mpi_p, 0, MPI_COMM_WORLD, &
ierr)
call MPI_GATHER(minval(x_cb), 1, mpi_p, &
ext_temp, 1, mpi_p, 0, &
MPI_COMM_WORLD, ierr)
if (proc_rank == 0) spatial_extents(1, :) = ext_temp

! Maximum spatial extent in the x-direction
call MPI_GATHERV(maxval(x_cb), 1, mpi_p, &
spatial_extents(2, 0), recvcounts, 4*displs, &
mpi_p, 0, MPI_COMM_WORLD, &
ierr)
call MPI_GATHER(maxval(x_cb), 1, mpi_p, &
ext_temp, 1, mpi_p, 0, &
MPI_COMM_WORLD, ierr)
if (proc_rank == 0) spatial_extents(2, :) = ext_temp
end if

#endif
Expand Down Expand Up @@ -339,16 +343,35 @@ contains

#ifdef MFC_MPI
integer :: ierr !< Generic flag used to identify and report MPI errors

! Minimum flow variable extent
call MPI_GATHERV(minval(q_sf), 1, mpi_p, &
data_extents(1, 0), recvcounts, 2*displs, &
mpi_p, 0, MPI_COMM_WORLD, ierr)

! Maximum flow variable extent
call MPI_GATHERV(maxval(q_sf), 1, mpi_p, &
data_extents(2, 0), recvcounts, 2*displs, &
mpi_p, 0, MPI_COMM_WORLD, ierr)
real(wp) :: ext_temp(0:num_procs - 1)

if (n > 0) then
! Multi-D: recvcounts = 1, so strided MPI_GATHERV works correctly
! Minimum flow variable extent
call MPI_GATHERV(minval(q_sf), 1, mpi_p, &
data_extents(1, 0), recvcounts, 2*displs, &
mpi_p, 0, MPI_COMM_WORLD, ierr)

! Maximum flow variable extent
call MPI_GATHERV(maxval(q_sf), 1, mpi_p, &
data_extents(2, 0), recvcounts, 2*displs, &
mpi_p, 0, MPI_COMM_WORLD, ierr)
else
! 1D: recvcounts/displs are sized for grid defragmentation
! (m+1 per rank), not for scalar gathers. Use MPI_GATHER instead.

! Minimum flow variable extent
call MPI_GATHER(minval(q_sf), 1, mpi_p, &
ext_temp, 1, mpi_p, 0, &
MPI_COMM_WORLD, ierr)
if (proc_rank == 0) data_extents(1, :) = ext_temp

! Maximum flow variable extent
call MPI_GATHER(maxval(q_sf), 1, mpi_p, &
ext_temp, 1, mpi_p, 0, &
MPI_COMM_WORLD, ierr)
if (proc_rank == 0) data_extents(2, :) = ext_temp
end if

#endif

Expand Down
Loading