no code implementations • 24 Feb 2021 • Rahim Mammadli, Marija Selakovic, Felix Wolf, Michael Pradel
Applying the transformations that our model deems most favorable prior to compilation yields an average speedup of 1. 14x.