When running Fast LAS with multithreading enabled, the results may vary between executions, even when using the same input data and fixed random seeds.
This suggests the presence of non-deterministic behavior, likely due to parallel execution (e.g., race conditions, non-deterministic ordering of operations, or floating-point accumulation differences across threads).
Is there any way to control or enforce deterministic behavior when using multithreading (e.g., configuration flags, deterministic mode, or recommended practices)?
When running Fast LAS with multithreading enabled, the results may vary between executions, even when using the same input data and fixed random seeds.
This suggests the presence of non-deterministic behavior, likely due to parallel execution (e.g., race conditions, non-deterministic ordering of operations, or floating-point accumulation differences across threads).
Is there any way to control or enforce deterministic behavior when using multithreading (e.g., configuration flags, deterministic mode, or recommended practices)?