DROPS

Document

DOI: 10.4230/LIPIcs.ECRTS.2023.4

Memory Latency Distribution-Driven Regulation for Temporal Isolation in MPSoCs

Authors: Ahsan Saeed, Denis Hoornaert, Dakshina Dasari, Dirk Ziegenbein, Daniel Mueller-Gritschneder, Ulf Schlichtmann, Andreas Gerstlauer, and Renato Mancuso

Published in: LIPIcs, Volume 262, 35th Euromicro Conference on Real-Time Systems (ECRTS 2023)

Abstract

Temporal isolation is one of the most significant challenges that must be addressed before Multi-Processor Systems-on-Chip (MPSoCs) can be widely adopted in mixed-criticality systems with both time-sensitive real-time (RT) applications and performance-oriented non-real-time (NRT) applications. Specifically, the main memory subsystem is one of the most prevalent causes of interference, performance degradation and loss of isolation. Existing memory bandwidth regulation mechanisms use static, dynamic, or predictive DRAM bandwidth management techniques to restore the execution time of an application under contention as close as possible to the execution time in isolation. In this paper, we propose a novel distribution-driven regulation whose goal is to achieve a timeliness objective formulated as a constraint on the probability of meeting a certain target execution time for the RT applications. Using existing interconnect-level Performance Monitoring Units (PMU), we can observe the Cumulative Distribution Function (CDF) of the per-request memory latency. Regulation is then triggered to enforce first-order stochastical dominance with respect to a desired reference. Consequently, it is possible to enforce that the overall observed execution time random variable is dominated by the reference execution time. The mechanism requires no prior information of the contending application and treats the DRAM subsystem as a black box. We provide a full-stack implementation of our mechanism on a Commercial Off-The-Shelf (COTS) platform (Xilinx Ultrascale+ MPSoC), evaluate it using real and synthetic benchmarks, experimentally validate that the timeliness objectives are met for the RT applications, and demonstrate that it is able to provide 2.2x more overall throughput for NRT applications compared to DRAM bandwidth management-based regulation approaches.

Cite as

Ahsan Saeed, Denis Hoornaert, Dakshina Dasari, Dirk Ziegenbein, Daniel Mueller-Gritschneder, Ulf Schlichtmann, Andreas Gerstlauer, and Renato Mancuso. Memory Latency Distribution-Driven Regulation for Temporal Isolation in MPSoCs. In 35th Euromicro Conference on Real-Time Systems (ECRTS 2023). Leibniz International Proceedings in Informatics (LIPIcs), Volume 262, pp. 4:1-4:23, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2023)

Copy BibTex To Clipboard

@InProceedings{saeed_et_al:LIPIcs.ECRTS.2023.4,
  author =	{Saeed, Ahsan and Hoornaert, Denis and Dasari, Dakshina and Ziegenbein, Dirk and Mueller-Gritschneder, Daniel and Schlichtmann, Ulf and Gerstlauer, Andreas and Mancuso, Renato},
  title =	{{Memory Latency Distribution-Driven Regulation for Temporal Isolation in MPSoCs}},
  booktitle =	{35th Euromicro Conference on Real-Time Systems (ECRTS 2023)},
  pages =	{4:1--4:23},
  series =	{Leibniz International Proceedings in Informatics (LIPIcs)},
  ISBN =	{978-3-95977-280-8},
  ISSN =	{1868-8969},
  year =	{2023},
  volume =	{262},
  editor =	{Papadopoulos, Alessandro V.},
  publisher =	{Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
  address =	{Dagstuhl, Germany},
  URL =		{https://drops.dagstuhl.de/entities/document/10.4230/LIPIcs.ECRTS.2023.4},
  URN =		{urn:nbn:de:0030-drops-180339},
  doi =		{10.4230/LIPIcs.ECRTS.2023.4},
  annote =	{Keywords: temporal isolation, memory latency, real-time system, multi-core}
}

Document

DOI: 10.4230/DagRep.11.6.147

Approximate Systems (Dagstuhl Seminar 21302)

Authors: Eva Darulova, Babak Falsafi, Andreas Gerstlauer, and Phillip Stanley-Marbell

Published in: Dagstuhl Reports, Volume 11, Issue 6 (2021)

Abstract

This report summarizes the presentations and discussion sessions at the Dagstuhl Seminar 21302 "Approximate Systems" that took place during July 25 - 30, 2021. Due to COVID, the seminar was held in a hybrid fashion, with around 1/3 of the attendees on-site and the remaining ones online. The seminar discussed advances and open challenges in applying approximate computing techniques across the stack and across different application domains, and we hope that this report can provide a useful resource also for other researchers.

Cite as

Eva Darulova, Babak Falsafi, Andreas Gerstlauer, and Phillip Stanley-Marbell. Approximate Systems (Dagstuhl Seminar 21302). In Dagstuhl Reports, Volume 11, Issue 6, pp. 147-163, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2021)

Copy BibTex To Clipboard

@Article{darulova_et_al:DagRep.11.6.147,
  author =	{Darulova, Eva and Falsafi, Babak and Gerstlauer, Andreas and Stanley-Marbell, Phillip},
  title =	{{Approximate Systems (Dagstuhl Seminar 21302)}},
  pages =	{147--163},
  journal =	{Dagstuhl Reports},
  ISSN =	{2192-5283},
  year =	{2021},
  volume =	{11},
  number =	{6},
  editor =	{Darulova, Eva and Falsafi, Babak and Gerstlauer, Andreas and Stanley-Marbell, Phillip},
  publisher =	{Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
  address =	{Dagstuhl, Germany},
  URL =		{https://drops.dagstuhl.de/entities/document/10.4230/DagRep.11.6.147},
  URN =		{urn:nbn:de:0030-drops-155836},
  doi =		{10.4230/DagRep.11.6.147},
  annote =	{Keywords: approximate computing, energy-efficient computing, pareto optimization}
}

Search Results

Documents authored by Gerstlauer, Andreas

Memory Latency Distribution-Driven Regulation for Temporal Isolation in MPSoCs

Abstract

Cite as

Approximate Systems (Dagstuhl Seminar 21302)

Abstract

Cite as

Thanks for your feedback!

Could not send message