We are developping our own tools. What are some strategies to accelerate EPFD validation run calculations?

Current implementations are using parallel computations. With 2 different approaches – parallelizing calculation by time steps, and by different runs. Latter looks more efficient. There were also some attempts to use GPU computing, but it is somewhat limited because GPU shared memory is very limited maximum only 64 kb which is very low to contain the mask.
But potentially it can be investigated for up-link calculations where EIRP-mask are normally small in size.
Also, you can use some internal optimizations like for finding alpha-angle which can be represented as table.