Custom Query (964 matches)
Results (25 - 27 of 964)
| Ticket | Owner | Reporter | Resolution | Summary |
|---|---|---|---|---|
| #25 | Fixed | Declaration problem in cgradient.h | ||
| Description |
In file cgradient.h, subroutine read_state.F, the explicit-shape version of the declaration for argument s_t is real(r8), intent(inout) :: s_t(LBi:UBi,LBj:UBj,N(ng),NT(ng)) This has only 4 dimensions, whereas the corresponding assumed-shape declaration has 5: real(r8), intent(inout) :: s_t(LBi:,LBj:,:,:,:) I think the correct explicit-shape declaration is real(r8), intent(inout) :: s_t(LBi:UBi,LBj:UBj,N(ng),2,NT(ng)) Corrected file attached. |
|||
| #26 | Fixed | ROMS fails to initialize in parallel with point sources/sink | ||
| Description |
I ran the LATTE model for several cases with ROMS 3.0 changeset 56. Compiled on moby using pgf90. If I set tiles to 2x2 (or 1x4) the program freezes on initialization(at runtime). This does not happen with a 1x1 tile or if I compile in serial. Also, the upwelling case compiles and initializes fine at runtime. Once UV_SOURCE and TS_SOURCE are undefined there is no apparent problem running in parallel. |
|||
| #28 | WorksForMe | Application ADRIA02 and river | ||
| Description |
The application ADRIA02 of ROMS 3.0 (with no modification whatsoever) does not run when using MPI and NtileI*NTileJ > 1. ROMS remain stuck forever at
It runs correctly with NtileI=NtileJ = 1. Disabling UV_PSOURCE and TS_PSOURCE gives a correct behavior. This problem occured on a AltiX system with ifort 8.1 I am ready to help as much as possible if the bug is not reproducible.
|
|||
