Correlation of Performance Optimizations and Energy Consumption for Stencil-Based Application on Intel Xeon Scalable Processors

Abstract

This article provides a comprehensive study of the impact of performance optimizations on the energy efficiency of a real-world CFD application called MPDATA, as well as an insightful analysis of performance-energy interaction of these optimizations with the underlying hardware that represents the first generation of Intel Xeon Scalable processors. Considering the MPDATA iterative application as a use case, we explore the fundamentals of energy and performance analysis for a memory-bound application when exposed to a set of optimization steps that increase the application performance, by improving the operational intensity of code and utilizing resources more efficiently. It is shown that for memory-bound applications, optimizing toward high performance could be a powerful strategy for improving the energy efficiency as well. In fact, for the considered performance optimizations, the energy gain is correlated with the performance gain but with varying degrees. As a result, these optimizations allow improving both performance and energy consumption radically, up to about 10.9 and 8.8 times, respectively. The impact of the Intel AVX-512 SIMD extension on the energy consumption and performance is demonstrated. Also, we discover limitations on the usability of CPU frequency scaling as a tool for balancing energy savings with admissible performance losses.

Architectural Adaptation and Performance-Energy Optimization for CFD Application on AMD EPYC Rome

“…As a result, the relatively low operational intensity of each MPDATA kernel [4] is not high enough to efficiently utilize the resources of modern processors. In our works [4], [11], [21], [35], [38], a set of optimizations was developed to exploit resources of multicore ccNUMA/SMP systems more efficiently. The resulting parallelization methodology consists of the following parametric optimization steps: [21] -this step explores spatial blocking across the different kernels, employing overlapped tiling with redundant computations, while all kernels are grouped into five packages using loop fusion.…”

“…Among these systems were 2-socket servers with Intel Xeon CPUs based on Skylake SP, Broadwell, and Haswell architectures. For example, for a platform built with 28-core Intel Platinum 8180 CPUs, the proposed adaptation accelerates the MPDATA application more than 10 times [11] compared to the basic version of code.…”

“…Finally, details of vectorizing computation within cores are determined. For Intel 8180 and 8280 CPUs, the AVX-512 SIMD extension is applied, allowing to increase the performance on the 2-socket 8180 CPUs about 2.8 times [11]…”

“…Besides the tendency of energy reduction by re-engineering the hardware, another trend is observed with great potential for energy savings realized by transforming and completely rethinking the algorithms [32] and software [33] dedicated for HPC systems. This work belongs mainly to the second trend, and partly follows our previous paper [11] that studied the impact of performance optimizations on the energy efficiency of MPDATA for the first generation of Intel Xeon Scalable processors. [34].…”

“…It is expected that the resulting code can adapt to the parameters of hardware components and their interaction with the proposed parametric optimizations. Furthermore, based on results achieved in our work [11], this study investigates the correlation between performance and energy efficiency for a ccNUMA platform powered by two top-of-the-line AMD Rome 7742 CPUs, each with 64 cores. The hardware-based technique [12] provides the accuracy and reliability of energy/power measurements with the Yokogawa WT310 power meter [13].…”