Approximate computing is a new paradigm for energy efficient design, based on the idea of designing digital systems that trade off computational accuracy for energy consumption. The paradigm can be applied to different units (i.e. internal units of the CPU, floating point coprocessors, memories). Considering the memory subsystem, approximate memories are physical memories where circuit-level or architecture-level techniques are implemented in order to reduce energy at the expense of errors occurring in bit cells. Supporting approximate memories at operating system level is required for managing them efficiently and for allowing user level applications to use it directly, but its implementation is subject to specific requirements and constraints, sometimes architecture dependent. In this paper we describe the introduction of approximate memory support on ARM architectures, which are widely adopted in low power embedded systems. While Linux support for approximate memory has already been introduced for main allocators, porting it to ARM architectures required the introduction of its specific support in the Linux early allocators, that are a fundamental function of the Linux kernel startup phase, before instantiation of the main allocators.

Approximate memory support for Linux early allocators in ARM architectures / Stazi, G.; Mastrandrea, A.; Olivieri, M.; Menichelli, F.. - 573:9783030119720(2019), pp. 429-435. (Intervento presentato al convegno International Conference on Applications in Electronics Pervading Industry, Environment and Society, APPLEPIES 2018 tenutosi a Pisa, Italy) [10.1007/978-3-030-11973-7_51].

Approximate memory support for Linux early allocators in ARM architectures

Stazi G.;Mastrandrea A.;Olivieri M.;Menichelli F.
2019

Abstract

Approximate computing is a new paradigm for energy efficient design, based on the idea of designing digital systems that trade off computational accuracy for energy consumption. The paradigm can be applied to different units (i.e. internal units of the CPU, floating point coprocessors, memories). Considering the memory subsystem, approximate memories are physical memories where circuit-level or architecture-level techniques are implemented in order to reduce energy at the expense of errors occurring in bit cells. Supporting approximate memories at operating system level is required for managing them efficiently and for allowing user level applications to use it directly, but its implementation is subject to specific requirements and constraints, sometimes architecture dependent. In this paper we describe the introduction of approximate memory support on ARM architectures, which are widely adopted in low power embedded systems. While Linux support for approximate memory has already been introduced for main allocators, porting it to ARM architectures required the introduction of its specific support in the Linux early allocators, that are a fundamental function of the Linux kernel startup phase, before instantiation of the main allocators.
2019
International Conference on Applications in Electronics Pervading Industry, Environment and Society, APPLEPIES 2018
approximate computing; approximate memory; Linux OS; low power embedded systems
04 Pubblicazione in atti di convegno::04b Atto di convegno in volume
Approximate memory support for Linux early allocators in ARM architectures / Stazi, G.; Mastrandrea, A.; Olivieri, M.; Menichelli, F.. - 573:9783030119720(2019), pp. 429-435. (Intervento presentato al convegno International Conference on Applications in Electronics Pervading Industry, Environment and Society, APPLEPIES 2018 tenutosi a Pisa, Italy) [10.1007/978-3-030-11973-7_51].
File allegati a questo prodotto
File Dimensione Formato  
Stazi_Approximate_post-print_2018.pdf

accesso aperto

Tipologia: Documento in Post-print (versione successiva alla peer review e accettata per la pubblicazione)
Licenza: Tutti i diritti riservati (All rights reserved)
Dimensione 95.42 kB
Formato Adobe PDF
95.42 kB Adobe PDF

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11573/1291445
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 0
  • ???jsp.display-item.citation.isi??? ND
social impact