Cuttlefish: Library for Achieving Energy Efficiency in Multicore Parallel Programs

被引：6
作者：
Kumar, Sunil [1 ]
Gupta, Akshat [1 ]
Kumar, Vivek [1 ]
Bhalachandra, Sridutt [2 ]
机构：
[1] IIIT Delhi, Delhi, India
[2] Lawrence Berkeley Natl Lab, Berkeley, NJ USA
来源：
SC21: INTERNATIONAL CONFERENCE FOR HIGH PERFORMANCE COMPUTING, NETWORKING, STORAGE AND ANALYSIS | 2021年
基金：
美国国家科学基金会;
关键词：
Multicore parallelism; DVFS; UFS; energy efficiency; SYSTEM;
D O I：
10.1145/3458817.3476163
中图分类号：
TP3 [计算技术、计算机技术];
学科分类号：
0812 ;
摘要：
A low-cap power budget is challenging for exascale computing. Dynamic Voltage. and freqtrency Scaling (DNTS) and Uncore freqtrency Scaling (LIPS) are the two widely used techniques for limiting the IIPC application's energy footprint. However, existing approaches fail to provide a unified solution that can work with different types of parallel programming models and applications. This paper proposes ClatIefish, a programming model oblivious C/C++ library for achieving energy efficiency in rnulticore parallel programs running over Intel processors. An online profiler periodically profiles model specific registers to discover a running application's memory access pattern. Using a combination of DVFS and UFS, Cuttlefish then dynamically adapts the processor's core and uncore frequencies, thereby improving its energy efficiency. The evaluation on a 20-core Intel Xeon processor using a set of widely used OpenMP benchmarks, consisting of several irregular-tasking and work -sharing pragmas, achieves geometric mean energy savings of 19.4% with a 3.6% slowdown.
引用
收藏
页数：14
相关论文
共 53 条
[31]  
Li D., 2010, HYBRID MPI OPENMP PO, P1, DOI DOI 10.1109/IPDPS.2010.5470463
[32]  
llnl, About us
[33]  
LLNL, MSR SAFE
[34]  
LLNL, EXASCALE COMPUTING P
[35]  
M.I.T, 2010, CILK 5 4 6
[36]  
Maury M.C., 2006, Proceedings of the 20th annual international conference on Supercomputing, ICS '06, P157
[37]  
Olivier Stephen, 2006, Languages and compliers for parallel computing, P235
[38]  
Porterfield Allan., 2010, RCRTOOL DESIGN DOCUM
[39]  
Porterfield Allan, 2013, P 1 INT WORKSHOP ENE, DOI [10.1145/2536430.2536437, DOI 10.1145/2536430.2536437]
[40]   EEWA: Energy-EfficientWorkload-Aware Task Scheduling in Multi-core Architectures [J].
Chen, Quan ;
Zheng, Long ;
Guo, Minyi ;
Huang, Zhiyi .
PROCEEDINGS OF 2014 IEEE INTERNATIONAL PARALLEL & DISTRIBUTED PROCESSING SYMPOSIUM WORKSHOPS (IPDPSW), 2014, :643-652
123456