https://hgpu.org/?p=17277
Speedup and Parallelization Models for Energy-Efficient Many-Core Systems Using Performance Counters