Nitro: A Framework for Adaptive Code Variant Tuning

• Published in: Proceedings - IEEE International Parallel and Distributed Processing Symposium pp. 501 - 512
• Main Authors: Muralidharan, Saurav, Shantharam, Manu, Hall, Mary, Garland, Michael, Catanzaro, Bryan
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.05.2014
• Subjects: Autotuning; GPUs; Libraries; Optimization; Performance Optimization; Sparse matrices; Training; Tuners; Vectors
• ISBN: 1479937991; 9781479937998
• ISSN: 1530-2075
• DOI: 10.1109/IPDPS.2014.59

Autotuning systems intelligently navigate a search space of possible implementations of a computation to find the implementation(s) that best meets a specific optimization criteria, usually performance. This paper describes Nitro, a programmer-directed auto tuning framework that facilitates tuning of code variants, or alternative implementations of the same computation. Nitro provides a library interface that permits programmers to express code variants along with meta-information that aids the system in selecting among the set of variants at run time. Machine learning is employed to build a model through training on this meta-information, so that when a new input is presented, Nitro can consult the model to select the appropriate variant. In experiments with five real-world irregular GPU benchmarks from sparse numerical methods, graph computations and sorting, Nitro-tuned variants achieve over 93% of the performance of variants selected through exhaustive search. Further, we describe optimizations and heuristics in Nitro that substantially reduce training time and other overheads.