TLB Shootdown Mitigation for Low-Power Many-Core Servers with L1 Virtual Caches

Binh Pham, Derek Hower, Abhishek Bhattacharjee, Trey Cain

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

Power efficiency has become one of the most important design constraints for high-performance systems. In this paper, we revisit the design of low-power virtually-addressed caches. While virtually-addressed caches enable significant power savings by obviating the need for Translation Lookaside Buffer (TLB) lookups, they suffer from several challenging design issues that curtail their widespread commercial adoption. We focus on one of these challenges-cache flushes due to virtual page remappings. We use detailed studies on an ARM many-core server to show that this problem degrades performance by up to 25 percent for a mix of multi-programmed and multi-threaded workloads. Interestingly, we observe that many of these flushes are spurious, and caused by an indiscriminate invalidation broadcast on ARM architecture. In response, we propose a low-overhead and readily implementable hardware mechanism using bloom filters to reduce spurious invalidations and mitigate their ill effects.

Original languageEnglish (US)
Pages (from-to)17-20
Number of pages4
JournalIEEE Computer Architecture Letters
Volume17
Issue number1
DOIs
StatePublished - Jan 1 2018

Fingerprint

Servers
Hardware

All Science Journal Classification (ASJC) codes

  • Hardware and Architecture

Keywords

  • TLB
  • Virtual Cache
  • multicores
  • multiprogramming
  • multithreading
  • virtual memory

Cite this

Pham, Binh ; Hower, Derek ; Bhattacharjee, Abhishek ; Cain, Trey. / TLB Shootdown Mitigation for Low-Power Many-Core Servers with L1 Virtual Caches. In: IEEE Computer Architecture Letters. 2018 ; Vol. 17, No. 1. pp. 17-20.
@article{15a415c3c2764369ab632d5218f59eb9,
title = "TLB Shootdown Mitigation for Low-Power Many-Core Servers with L1 Virtual Caches",
abstract = "Power efficiency has become one of the most important design constraints for high-performance systems. In this paper, we revisit the design of low-power virtually-addressed caches. While virtually-addressed caches enable significant power savings by obviating the need for Translation Lookaside Buffer (TLB) lookups, they suffer from several challenging design issues that curtail their widespread commercial adoption. We focus on one of these challenges-cache flushes due to virtual page remappings. We use detailed studies on an ARM many-core server to show that this problem degrades performance by up to 25 percent for a mix of multi-programmed and multi-threaded workloads. Interestingly, we observe that many of these flushes are spurious, and caused by an indiscriminate invalidation broadcast on ARM architecture. In response, we propose a low-overhead and readily implementable hardware mechanism using bloom filters to reduce spurious invalidations and mitigate their ill effects.",
keywords = "TLB, Virtual Cache, multicores, multiprogramming, multithreading, virtual memory",
author = "Binh Pham and Derek Hower and Abhishek Bhattacharjee and Trey Cain",
year = "2018",
month = "1",
day = "1",
doi = "10.1109/LCA.2017.2712140",
language = "English (US)",
volume = "17",
pages = "17--20",
journal = "IEEE Computer Architecture Letters",
issn = "1556-6056",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "1",

}

TLB Shootdown Mitigation for Low-Power Many-Core Servers with L1 Virtual Caches. / Pham, Binh; Hower, Derek; Bhattacharjee, Abhishek; Cain, Trey.

In: IEEE Computer Architecture Letters, Vol. 17, No. 1, 01.01.2018, p. 17-20.

Research output: Contribution to journalArticle

TY - JOUR

T1 - TLB Shootdown Mitigation for Low-Power Many-Core Servers with L1 Virtual Caches

AU - Pham, Binh

AU - Hower, Derek

AU - Bhattacharjee, Abhishek

AU - Cain, Trey

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Power efficiency has become one of the most important design constraints for high-performance systems. In this paper, we revisit the design of low-power virtually-addressed caches. While virtually-addressed caches enable significant power savings by obviating the need for Translation Lookaside Buffer (TLB) lookups, they suffer from several challenging design issues that curtail their widespread commercial adoption. We focus on one of these challenges-cache flushes due to virtual page remappings. We use detailed studies on an ARM many-core server to show that this problem degrades performance by up to 25 percent for a mix of multi-programmed and multi-threaded workloads. Interestingly, we observe that many of these flushes are spurious, and caused by an indiscriminate invalidation broadcast on ARM architecture. In response, we propose a low-overhead and readily implementable hardware mechanism using bloom filters to reduce spurious invalidations and mitigate their ill effects.

AB - Power efficiency has become one of the most important design constraints for high-performance systems. In this paper, we revisit the design of low-power virtually-addressed caches. While virtually-addressed caches enable significant power savings by obviating the need for Translation Lookaside Buffer (TLB) lookups, they suffer from several challenging design issues that curtail their widespread commercial adoption. We focus on one of these challenges-cache flushes due to virtual page remappings. We use detailed studies on an ARM many-core server to show that this problem degrades performance by up to 25 percent for a mix of multi-programmed and multi-threaded workloads. Interestingly, we observe that many of these flushes are spurious, and caused by an indiscriminate invalidation broadcast on ARM architecture. In response, we propose a low-overhead and readily implementable hardware mechanism using bloom filters to reduce spurious invalidations and mitigate their ill effects.

KW - TLB

KW - Virtual Cache

KW - multicores

KW - multiprogramming

KW - multithreading

KW - virtual memory

UR - http://www.scopus.com/inward/record.url?scp=85041706933&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85041706933&partnerID=8YFLogxK

U2 - 10.1109/LCA.2017.2712140

DO - 10.1109/LCA.2017.2712140

M3 - Article

AN - SCOPUS:85041706933

VL - 17

SP - 17

EP - 20

JO - IEEE Computer Architecture Letters

JF - IEEE Computer Architecture Letters

SN - 1556-6056

IS - 1

ER -