Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
Understanding Energy Behaviors of Thread Management Constructs
Search
Gustavo Pinto
January 14, 2018
0
65
Understanding Energy Behaviors of Thread Management Constructs
Gustavo Pinto
January 14, 2018
Tweet
Share
More Decks by Gustavo Pinto
See All by Gustavo Pinto
Apoiando pessoas programadoras com deficiência: Lições aprendidas na Zup Innovation
gustavopinto
0
34
5 coisas que todo dev deveria saber Para criar aplicações com GenAI
gustavopinto
0
39
Large Language Models for Education: Grading Open-Ended Questions Using ChatGPT
gustavopinto
0
54
Cognitive Driven Development: A Research Agenda
gustavopinto
0
37
The Infinite Academic Game
gustavopinto
0
32
[ESEM 2022] To What Extent Cognitive Driven Development Improves Code Readability
gustavopinto
0
74
Caminhos e desafios para a pesquisa em Computação (ou como se manter produtivo) na Região Norte
gustavopinto
2
270
How Open is the SBES PC Community
gustavopinto
0
70
REST or GraphQL? A Performance Comparative Study
gustavopinto
0
210
Featured
See All Featured
The Success of Rails: Ensuring Growth for the Next 100 Years
eileencodes
30
6k
jQuery: Nuts, Bolts and Bling
dougneiner
59
7.1k
CSS Pre-Processors: Stylus, Less & Sass
bermonpainter
352
28k
Mobile First: as difficult as doing things right
swwweet
216
8.6k
Fantastic passwords and where to find them - at NoRuKo
philnash
36
2.5k
Building a Modern Day E-commerce SEO Strategy
aleyda
16
6.4k
"I'm Feeling Lucky" - Building Great Search Experiences for Today's Users (#IAC19)
danielanewman
220
21k
Creatively Recalculating Your Daily Design Routine
revolveconf
209
11k
Build The Right Thing And Hit Your Dates
maggiecrowley
23
2k
Keith and Marios Guide to Fast Websites
keithpitt
408
22k
Writing Fast Ruby
sferik
620
60k
From Idea to $5000 a Month in 5 Months
shpigford
377
45k
Transcript
Understanding Energy Behaviors of Thread Management Constructs Gustavo Pinto1 Fernando
Castor1 David Liu2 {ghlp, castor}@cin.ufpe.br1
[email protected]
2 • First, the proliferation of multicore CPUs • Second,
the prevalence of multi-threaded programs Motivation
3 • First, more cores more power consumed • Second,
little is known about energy behaviors of multi-threaded program on the application and programming language level The Problem
4 1. programming abstractions of thread management on energy efficiency
2. programmer choices of thread management on energy efficiency This Talk for Java multi-threaded programs
5 • Explicit threading (the Thread-style): Using the java.lang.Thread class
• Thread pooling (the Executor-style): Using the java.util.concurrent.Executor framework • Working Stealing (the ForkJoin-style): Using the java.util.concurrent.ForkJoin framework Thread management constructs
6 • Embarrassingly parallel: spectralnorm, sunflow, n-queens • Leaning parallel:
xalan, knucleotide, tomcat • Leaning serial: mandelbrot, largestImage • Embarrassingly serial: h2 Benchmarks
7 • Embarrassingly parallel: spectralnorm, sunflow, n-queens • Leaning parallel:
xalan, knucleotide, tomcat • Leaning serial: mandelbrot, largestImage • Embarrassingly serial: h2 Benchmarks Micro-benchmarks DaCapo benchmarks
8 Experimental Environment A 2×16-core AMD CPUs, running Debian Linux,
64GB of memory, JDK version 1.7.0 11, build 21, “ondemand” governor
9 Experimental Environment A 2×16-core AMD CPUs, running Debian Linux,
64GB of memory, JDK version 1.7.0 11, build 21, “ondemand” governor
10 Experimental Environment A 2×16-core AMD CPUs, running Debian Linux,
64GB of memory, JDK version 1.7.0 11, build 21.
11 Experimental Environment A 2×16-core AMD CPUs, running Debian Linux,
64GB of DDR3 1600 memory, and JDK version 1.7.0 11, build 21.
Energy Consumption When Varying the Number of Threads 12
13 The Λ Curve
14 The Λ Curve
15 The Λ Curve
16 More cores idle CPU frequency at a lower level
The Λ Curve
17 More cores idle CPU frequency at a lower level
More threads used, program completes sooner The greater the ratio between speedup and power, the steeper the \ The Λ Curve
Which programming style should I use? 18 ? ?
Overpopulating Cores with Threads 19
Overpopulating Cores with Threads 20
Faster ≠ Greener 21
Copying vs Sharing 22
Copying vs Sharing 23 Copying
Copying vs Sharing 24 Copying Sharing
Copying vs Sharing 25 Copying Sharing ±15% of energy savings!
26 Copy-Fork
27 After Before Copy-Fork
28 After Before Copy-Fork Copy/Fork/Copy/Fork/…
29 After Before Copy-Fork Copy/Fork/Copy/Fork/…
30 After Before Copy-Fork Copy/Fork/Copy/Fork/… Copy/../Copy/Fork/…/Fork
31 After Before Copy-Fork ±10% of energy savings! Copy/Fork/Copy/Fork/… Copy/../Copy/Fork/…/Fork
Data Size 32
Data Size 33 For most of the cases, energy consumption
is linear to data size!
34
35
36
37
Understanding Energy Behaviors of Thread Management Constructs Gustavo Pinto1 Fernando
Castor1 David Liu2 {ghlp, castor}@cin.ufpe.br1
[email protected]