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 3 Algorithm results for about ten compute nodes.
 Benchmark Type $N$ $Δ?N*$ $f*$ Performance boost (%) Power saving (%) Energy saving (%) sp.D.169 Memory-bound 9 2 1.2 13.98 2.75 16.22 RandomAccess.160 Memory-bound 8 1 1.2 9.80 0.35 11.27 mg.D.128 Memory-bound 7 1 1.9 18.63 0.40 18.73 STREAM.160 Memory-bound 8 1 1.7 10.19 2.42 8.84 DGEMM.160 Memory-bound 8 2 1.5 18.82 2.12 19.85 cg.D.128 Memory-bound 7 2 1.2 -3.34 2.90 -0.29 bt.D.169 Memory-bound 9 1 1.8 -1.98 4.27 2.26 lu.D.128 Memory-bound 7 0 2.6 0 0 0 ft.D.128 CPU-bound 7 0 2.6 0 0 0 FFT.160 CPU-bound 8 0 2.6 0 0 0 PRANTS.160 CPU-bound 8 0 2.6 0 0 0 ep.D.128 CPU-bound 7 0 2.6 0 0 0