The underlying implementation is relatively slow (not compared to the RH15d implementation), and recreates the entire thread pool which is very expensive (and shouldn't be necessary now due to the change to the use of a View on the atom object which should automatically update).
Maybe see if we can use the thread pool (if already present) to accelerate the calculation of these terms.
|
cpdef update_hprd_coeffs(self): |
The underlying implementation is relatively slow (not compared to the RH15d implementation), and recreates the entire thread pool which is very expensive (and shouldn't be necessary now due to the change to the use of a View on the atom object which should automatically update).
Maybe see if we can use the thread pool (if already present) to accelerate the calculation of these terms.
Lightweaver/Source/LwMiddleLayer.pyx
Line 3632 in fd8045e