File-Per-Process IB markers - #1802
Conversation
|
Claude Code Review Head SHA: f0f2b3c Files changed:
Findings:
|
Lines of Code
|
|
@sbryngelson Do you still want us to keep PRs as draft when they are done? This and the restart changes would both be useful for me to get merged. So it would be helpful to have them bumped up in terms of priority for merge if possible |
|
noted. i'll prioritize it and move it out of draft soon enough |
There was a problem hiding this comment.
🟡 Changes recommended
The post-process source fails to compile without MPI because t_step_string is conditionally declared but unconditionally referenced.
Once you've addressed the issues Copilot identified, you can request another Copilot review.
Pull request overview
Adds file-per-process output and post-processing support for immersed-boundary marker data.
Changes:
- Writes rank-local IB marker files alongside timestep data.
- Reads rank-local marker files during post-processing.
File summaries
| File | Description |
|---|---|
src/simulation/m_data_output.fpp |
Writes per-rank IB marker files. |
src/post_process/m_data_input.f90 |
Reads per-rank IB marker files. |
Review details
- Files reviewed: 2/2 changed files
- Comments generated: 1
- Review effort level: Balanced
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
| character(len=*), intent(in) :: file_loc_base | ||
| integer, intent(in), optional :: t_step | ||
| character(LEN=path_len + 2*name_len) :: file_loc | ||
| logical :: file_exist | ||
| integer :: ifile, ierr, data_size | ||
|
|
||
| #ifdef MFC_MPI | ||
| integer, dimension(MPI_STATUS_SIZE) :: status | ||
| integer(KIND=MPI_OFFSET_KIND) :: disp | ||
| integer(KIND=MPI_OFFSET_KIND) :: m_MOK, n_MOK, p_MOK, MOK, WP_MOK, var_MOK | ||
| integer :: save_index | ||
| character(len=10) :: t_step_string | ||
| #endif |
For some reason, I never added file-per-process for IB markers, and it is a slowdown constaint for some of the IB cases we are running. Adding this here. I tested this on the
2D_mibm_shock_cylindercase on 2 ranks and verified the visual output was correct.Even though this is marked as "draft", it is ready for merge.