Would be interesting to see some performance benchmarks in comparison to previous iterations.