More Bang for Your Buck: Boosting Performance with Capped Power Consumption
Juan Chen,Xinxin Qi,Feihao Wu,Jianbin Fang,Yong Dong,Yuan Yuan,Zheng Wang,Keqin Li
Table 5 Algorithm results for about 30 compute nodes.
 Benchmark Type $N$ $Δ?N*$ $f*$ Performance boost (%) Power saving (%) Energy saving (%) sp.E.529 Memory-bound 27 4 1.9 10.36 8.02 16.23 RandomAccess.512 Memory-bound 26 1 1.2 21.11 11.19 30.37 mg.E.512 Memory-bound 26 5 1.8 13.00 8.72 20.45 STREAM.512 Memory-bound 26 3 1.9 13.70 0.35 14.73 DGEMM.512 Memory-bound 26 4 2.2 10.24 -1.21 8.24 cg.E.512 Memory-bound 26 3 1.9 3.14 11.18 13.25 bt.E.529 Memory-bound 27 2 2.0 17.31 11.20 24.50 lu.E.529 Memory-bound 27 0 2.6 0 0 0 ft.E.512 CPU-bound 26 0 2.6 0 0 0 FFT.512 CPU-bound 26 0 2.6 0 0 0 PRANTS.512 CPU-bound 26 0 2.6 0 0 0 ep.E.529 CPU-bound 27 0 2.6 0 0 0