Closed vektordev closed 8 years ago
Rather than profiling, I just used a different library (async) that can traverse over Trees. Additionally, a mutex provides a limit on the total number of concurrent ghc threads. This limits memory usage a bit. CPU usage is good, so the parallelization must've worked.
Currently, parallelization is a bit bad. Apparently, the bodged together parallelization over a Tree of jobs is not working properly. Profiling needed.