Restore I/O in x3div #25
Conversation
This "closes" the xdmf file. Also the code was starting at timestep 1 (this should be the value at the END of the first timestep)
| starts(1) = decomp%zst(1)-1 | ||
| starts(2) = decomp%zst(2)-1 | ||
| starts(3) = decomp%zst(3)-1 | ||
| endif |
There was a problem hiding this comment.
^ Maybe room for a method in decomp class ?
There was a problem hiding this comment.
Okay, I guess that's 2ddecomp code and you don't have a lot of control on it?
| disp = disp + sizes(1)*sizes(2)*sizes(3)*mytype_bytes | ||
| if (data_type == complex_type) then | ||
| disp = disp + sizes(1)*sizes(2)*sizes(3)*mytype_bytes | ||
| end if |
There was a problem hiding this comment.
This file looks the same as the previous one, except for the last 5 lines? How are they related? Do you need both?
There was a problem hiding this comment.
oldskool Fortran - I think we should leave it alone as it works. We can look at refactoring when we have merged onto the upstream decomp2d (no point fixing something we're going to discard)
|
|
||
| ndt = 0 | ||
|
|
||
| ioutput = 5 |
There was a problem hiding this comment.
Took me a bit of scrolling to understand what that variable was about, maybe a name like io_period?
| ! write(*,*)'===========================================================' | ||
| ! write(*,*)'Visu module requires ',real(memout*1e-9,4),'GB' | ||
| ! write(*,*)'===========================================================' | ||
| ! end if |
| #endif | ||
|
|
||
| if (nrank.eq.0) then | ||
| OPEN(newunit=ioxdmf,file="./data/"//pathname//"/"//filename//'-'//num//'.xdmf') |
There was a problem hiding this comment.
I'd extract writing xml to a thrid party library FoXy ? Or at a minimum into a separate module
| starts(3) = decomp%xst(3)-1 | ||
|
|
||
| call MPI_TYPE_CREATE_SUBARRAY(3, sizes, subsizes, starts, & | ||
| MPI_ORDER_FORTRAN, data_type, newtype, ierror) |
There was a problem hiding this comment.
Shall we add checks on the MPI error codes ?
| @@ -214,9 +220,11 @@ subroutine finalise_xcompact3d(flag) | |||
| call x3d_operator_z_data_finalize() | |||
| call var_finalize() | |||
|
|
|||
| call visu_finalise() | |||
There was a problem hiding this comment.
At some point, we will have to decide if we finalize or if we finalise :D
There was a problem hiding this comment.
I propose "fin" - gets around the argument of who spells correctly and sounds classy ;)
|
I have added some case-specific visualization. Seems to work on my workstation. Not tested with ADIOS2. |
|
I'm not sure it should support ADIOS2 (well) I used an older version of visu.f90 in interest of being able to use upstream 2decomp&fft |
No description provided.