Analytic performance model for parallel overlapping memory‐bound kernels
Open Access
- 13 January 2022
- journal article
- research article
- Published by Wiley in Concurrency and Computation: Practice and Experience
- Vol. 34 (10) , e6816
Abstract
Complex applications running on multicore processors show a rich performance phenomenology. The growing number of cores per ccNUMA domain complicates performance analysis of memory‐bound code since system noise, load imbalance, or task‐based programming models can lead to thread desynchronization. Hence, the simplifying assumption that all cores execute the same loop can not be upheld. Motivated by observations on plain and modified versions of the HPCG benchmark, we construct a performance model of execution of memory‐bound loop kernels. It can predict the memory bandwidth share per kernel on a memory contention domain depending on the number of active cores and which other workload the kernel is paired with. The only code features required are the single‐thread memory request fraction per kernel, which is directly related to the single‐thread memory bandwidth, and its saturated bandwidth. The former can either be measured directly or predicted using the Execution‐Cache‐Memory performance model. The computational intensity of the kernels and the detailed structure of the code is of no significance. We validate our model on Intel Broadwell, Intel Cascade Lake, and AMD Rome processors pairing various streaming and stencil kernels. The error in predicting the bandwidth share per kernel is less than 8%.Keywords
Funding Information
- Bundesministerium für Forschung, Technologie und Raumfahrt (Metacca, SeaSiTE)
This publication has 25 references indexed in Scilit:
- High-performance implementation of Chebyshev filter diagonalization for interior eigenvalue computationsJournal of Computational Physics, 2016
- Quantifying Performance Bottlenecks of Stencil Computations Using the Execution-Cache-Memory Model2015
- Reevaluating Data Stall Time with the Consideration of Data Access ConcurrencyJournal of Computer Science and Technology, 2015
- Idle waves in high-performance computingPhysical Review E, 2015
- Hiding global synchronization latency in the preconditioned Conjugate Gradient algorithmParallel Computing, 2014
- A Quantitative Analysis of OS Noise2011
- Introducing a Performance Model for Bandwidth-Limited Loop Kernels2010
- RooflineCommunications of the ACM, 2009
- System noise, OS clock ticks, and fine-grained parallel applications2005
- Microarchitecture Optimizations for Exploiting Memory-Level ParallelismACM SIGARCH Computer Architecture News, 2004