Closed dkyleward closed 8 years ago
Tested this using even just two threads and ran into problems. Permanently disabling multi-threading. This is a major blow to the run time of the model, and should be a major motivating factor for us to look to Python, which was designed for multi-threading from the start.
The original configuration of 15 threads and 10 GB of memory in the virtual machine led to errors in certain scenarios that didn't make sense. It did not affect every scenario. Turning off the parallel processing resolved these issues, but significantly increased model run time.
Test to see if 2, 4, or 8 threads can be run reliably to improve run time.