Automating the Application Data Placement in Hybrid Memory Systems
Authors:
- Harald ServatHSHarald Servat
- Antonio J. PenaAntonio J. Pena
- German LlortGLGerman Llort
- Estanislao MercadalEMEstanislao Mercadal
- Hans-Christian HoppeHHHans-Christian Hoppe
- Jesus LabartaJLJesus Labarta
- 1 September 2017
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 126-136
Abstract
Multi-tiered memory systems, such as those based on Intel ® Xeon Phi™ processors, are equipped with several memory tiers with different characteristics including, among others, capacity, access latency, bandwidth, energy consumption, and volatility. The proper distribution of the application data objects into the available memory layers is key to shorten the time- to-solution, but the way developers and end-users determine the most appropriate memory tier to place the application data objects has not been properly addressed to date. In this paper we present a novel methodology to build an extensible framework to automatically identify and place the application's most relevant memory objects into the Intel Xeon Phi fast on-package memory. Our proposal works on top of inproduction binaries by first exploring the application behavior and then substituting the dynamic memory allocations. This makes this proposal valuable even for end-users who do not have the possibility of modifying the application source code. We demonstrate the value of a framework based in our methodology for several relevant HPC applications using different allocation strategies to help end-users improve performance with minimal intervention. The results of our evaluation reveal that our proposal is able to identify the key objects to be promoted into fast on-package memory in order to optimize performance, leading to even surpassing hardware-based solutions.Keywords
This publication has 19 references indexed in Scilit:
- StructSlim: a lightweight profiler to guide structure splitting2016
- A data-oriented profiler to assist in data partitioning and distribution for heterogeneous memory in HPCParallel Computing, 2016
- Low-Overhead Detection of Memory Access Patterns and Their Time Evolution2015
- Dissecting On-Node Memory Access Performance: A Semantic Approach2014
- A non‐conforming multi‐element DGTD method for the simulation of human exposure to electromagnetic wavesInternational Journal of Numerical Modelling: Electronic Networks, Devices and Fields, 2013
- Quantifying the Potential Task-Based Dataflow Parallelism in MPI ApplicationsLecture Notes in Computer Science, 2011
- Pin2005
- An API for Runtime Code PatchingThe International Journal of High Performance Computing Applications, 2000
- MemSpy: analyzing memory system bottlenecks in programs1992
- The Nas Parallel BenchmarksThe International Journal of Supercomputing Applications, 1991