Performance/Energy Aware Optimization of Parallel Applications on GPUs Under Power Capping
Performance/Energy Aware Optimization of Parallel Applications on GPUs Under Power Capping
复制标题
功率上限下 GPU 上并行应用程序的性能/能源感知优化
DOI：
--
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
P. Czarnul
中科院分区：
文献类型：
--
作者：
Adam Krzywaniak;P. Czarnul
在本文中，我们介绍了将 NVIDIA GPU 可用的现代功率上限机制应用于 NAS 并行基准 BT、SP 和 LU 以及 cublasgemm 基准等基准的方法和结果，这些基准被广泛用于评估高性能计算系统的性能。具体来说，根据基准测试，各种功率上限配置最适合性能和能耗的所需权衡。我们提出两个：相同功率上限下的节能和性能下降以及标准化的性能-能耗乘积。重要的是，最佳配置通常不是平凡的，即在小于默认值且大于最小允许限制的功率上限下获得。我们已经对 Pascal 和 Turing 代的两种现代 GPU（分别是 NVIDIA GTX 1070 和 NVIDIA RTX 2080）进行了测试，因此结果对于许多应用程序来说非常有用，其配置文件类似于在基于现代 GPU 的系统上执行的基准测试。
In the paper we present an approach and results from application of the modern power capping mechanism available for NVIDIA GPUs to the benchmarks such as NAS Parallel Benchmarks BT, SP and LU as well as cublasgemm-benchmark which are widely used for assessment of high performance computing systems’ performance. Specifically, depending on the benchmarks, various power cap configurations are best for desired trade-off of performance and energy consumption. We present two: both energy savings and performance drops for same power caps as well as a normalized performance-energy consumption product. It is important that optimal configurations are often non-trivial i.e. are obtained for power caps smaller than default and larger than minimal allowed limits. Tests have been performed for two modern GPUs of Pascal and Turing generations i.e. NVIDIA GTX 1070 and NVIDIA RTX 2080 respectively and thus results can be useful for many applications with profiles similar to the benchmarks executed on modern GPU based systems.