Cleaned out the OpenMP statements
Removed all the OpenMP pragmas, OpenMP functions and header inclusions. Also replaced the associated atomic operations with normal operations.
There might have been some race condition problems in some routines of space.c where some non-atomic operations were performed in critical parts of the code. Not sure whether it was actually a problem.
Closes issue #20 (closed).
Merge request reports
Activity
Note that this change will cause a loss of scaling, as the serial fraction of the code has increased.
I've created a new issue to make a scheduler-based parallel sort of the particles, #43 (closed). I have some ideas on how to do this more or less elegantly.
mentioned in issue #43 (closed)
mentioned in commit 4cdc697d
Please register or sign in to reply