Clairvoyance is a compiler technique that targets the middle ground architectures, which strike a good balance between performance and energy efficiency and currently dominate the market for mobile, hand-held devices and high-end embedded systems. We show that these simple and more energy-efficient cores, equipped with the appropriate compile-time support, can reach new levels of performance for memory-bound applications. Our contribution is a compiler that reorders instructions to expose more fine-grain parallelism, which can then be easily exploited by the simple and more energy-efficient cores and significantly boost applications’ performance. Thus, we approach the problem of energy-efficiency by executing applications on more energy-efficient hardware and bringing back the lost performance.
To properly cite Clairvoyance, please use:
Tran, Kim-Anh, et al. "Clairvoyance: Look-ahead compile-time scheduling." 2017 IEEE/ACM International Symposium on Code Generation and Optimization (CGO). IEEE, 2017.