Model-Based Loop Perforation

In many applications there is a gap between the accuracy provided by the platform and the accuracy that is required by the application to produce good-enough results. Exploiting this gap specifically is the concept of Approximate Computing, where a small reduction in accuracy is traded for better performance or a reduction in energy consumption. We assess applications regarding their suitability to be approximated. We propose a novel approach for memory-aware perforation of GPU kernels. The technique is further optimized, and we show its applicability on embedded GPUs. In order to fully utilize the opportunities of our approach, we propose a novel framework for automatic loop nest approximation based on polyhedral compilation. Our approach generalizes state-of-the-art perforation techniques and introduces new multidimensional perforation schemes. Moreover, the approach is augmented with a reconstruction technique that significantly improves the accuracy of the results. As the transformation space is potentially large, we propose a pruning method to remove low-quality transformations.