parallelization block fixed setup (#122)

* parallelization block fixed setup using mincores = max(...) instead of min(...)

* adapt CHANGELOG.md accordingly

Signed-off-by: Marcel Müller <[email protected]>

---------

Signed-off-by: Marcel Müller <[email protected]>
Co-authored-by: Marcel Müller <[email protected]>

Author: lmseidler

CHANGELOG.md

@@ -25,6 +25,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - version string is now correctly formatted and printed
 - precision (# significant digits) of the coordinate files (`get_coord_str` and `get_xyz_str`) increased from 7 to 14
 - catch encoding errors when reading `Turbomole._run_opt` output files
+- bug in the parallelization, leading to a dead `mindlessgen` execution as a consequence of not allowing the required number of cores
 
 ## [0.5.0] - 2024-12-16
 ### Changed

src/mindlessgen/generator/main.py

@@ -124,7 +124,7 @@ def generator(config: ConfigManager) -> tuple[list[Molecule], int]:
     blocks = setup_blocks(
         num_cores,
         config.general.num_molecules,
-        min(config.refine.ncores, config.postprocess.ncores),
+        max(config.refine.ncores, config.postprocess.ncores),
     )
     blocks.sort(key=lambda x: x.ncores)
 

src/mindlessgen/prog/parallel.py

@@ -47,7 +47,7 @@ def setup_blocks(ncores: int, num_molecules: int, mincores: int) -> list[Block]:
     # Maximum and minimum number of parallel processes possible
     maxcores = ncores
     maxprocs = max(1, ncores // mincores)
-    minprocs = max(1, ncores // maxcores)
+    minprocs = 1
 
     # Distribute number of molecules among blocks
     # First (if possible) create the maximum number of parallel blocks (maxprocs) and distribute as many molecules as possible