introduction to java 8 parallelism...

Introduction to Java 8

Parallelism Frameworks

Douglas C. [email protected]

www.dre.vanderbilt.edu/~schmidt

Professor of Computer Science

Institute for Software

Integrated Systems

Vanderbilt University

Nashville, Tennessee, USA

mailto:[email protected]

2

• Recognize how Java 8 applies functional programming features to its parallelism frameworks

Learning Objectives in this Lesson

3

Learning Objectives in this Lesson• Recognize how Java 8 applies functional programming

features to its parallelism frameworks, e.g.

• Fork-join pools

4

• Recognize how Java 8 applies functional programming features to its parallelism frameworks, e.g.

• Fork-join pools

• Parallel streamsfilter(not(this::urlCached))

collect(toList())

map(this::downloadImage)

flatMap(this::applyFilters)


…

5


• Fork-join pools

• Parallel streams

• Completable futures


/page\ =

supplyAsync

(getStartPage())

/imgNum\ = /page\

.thenComposeAsync

(crawlHyperLinks

(page))

/imgNum\ = /page\

.thenApplyAsync

(countImages(page))

.thenApply(List::size)

/imgNum\.thenCombine(/imgNum\,

(imgNum, imgNum) ->

Integer::sum)

Completable futures also provide a reactive asynchrony programming model

6


• Know how these features are applied in several example case study apps

New &Improved


Socket

Socket

List of URLs to Download

…

List of Filters to Apply

Persistent

Data Store

7

Overview of Java 8 Parallelism Frameworks

8

• Java 7 introduced the fork-join pool

See www.infoq.com/interviews/doug-lea-fork-join


http://www.infoq.com/interviews/doug-lea-fork-join

9


• This framework provides high performance, fine-grained task execution for data parallelism

join joinjoin

Processsequentially

Processsequentially

Processsequentially

Processsequentially

DataSource1.1 DataSource1.2 DataSource2.1 DataSource2.2

DataSource1 DataSource2

DataSource

fork()

fork() fork()

See www.dre.vanderbilt.edu/~schmidt/PDF/DataParallelismInJava.pdf


http://www.dre.vanderbilt.edu/~schmidt/PDF/DataParallelismInJava.pdf

10See en.wikipedia.org/wiki/Divide_and_conquer_algorithm

Result solve(Problem problem) {

if (problem is small)

directly solve problem

else {

a. split problem into

independent parts

b. fork new sub-tasks

to solve each part

c. join all sub-tasks

d. compose result

from sub-results

}

}



• It supports parallel programming by solving problems via “divide & conquer”


https://en.wikipedia.org/wiki/Divide_and_conquer_algorithm

11



• It supports parallel programming by solving problems via “divide & conquer”

• Employs work-stealing to optimize multi-core processor performance

See gee.cs.oswego.edu/dl/papers/fj.pdf


http://gee.cs.oswego.edu/dl/papers/fj.pdf

12

• Java 8 adds two new parallelism frameworks related to functional programming

Additional Frameworks & Languages

Applications

Operating System Kernel

System Libraries

Java Execution Environment (e.g., JVM)

Threading & Synchronization Packages

Java

/JN

IC+

+/C

CSee www.ibm.com/developerworks/library/j-jvmc2


http://www.ibm.com/developerworks/library/j-jvmc2

13


1. Parallel streams filter(not(this::urlCached))

collect(toList())

…



See docs.oracle.com/javase/tutorial/collections/streams/parallelism.html


http://docs.oracle.com/javase/tutorial/collections/streams/parallelism.html

14


1. Parallel streams

• Partitions a stream into multiple substreams that run independently & combine into a “reduced” result

filter(not(this::urlCached))

collect(toList())

…




15


1. Parallel streams


• Chunks of data in the substreamscan be mapped to multiple threads(& cores)


collect(toList())

…




16


1. Parallel streams



• Leverage the common fork-join pool


collect(toList())

…




See dzone.com/articles/common-fork-join-pool-and-streams

https://dzone.com/articles/common-fork-join-pool-and-streams

17


1. Parallel streams



• Leverage the common fork-join pool


collect(toList())

…



Parallel streams provides fine-grained data parallelism functional programming


18


1. Parallel streams

2. Completable futures

See docs.oracle.com/javase/8/docs/api/java/util/concurrent/CompletableFuture.html

/page\ =

supplyAsync

(getStartPage())

/imgNum2\ = /page\

.thenComposeAsync

(crawlHyperLinks

(page))

/imgNum1\ = /page\

.thenApplyAsync

(countImages(page))


/imgNum1\.thenCombine(/imgNum2\,

(imgNum, imgNum) ->

Integer::sum)


https://docs.oracle.com/javase/8/docs/api/java/util/concurrent/CompletableFuture.html

19


1. Parallel streams


• Supports dependent actionsthat trigger upon completion of async operations

See docs.oracle.com/javase/8/docs/api/java/util/concurrent/CompletionStage.html

/page\ =

supplyAsync

(getStartPage())

/imgNum2\ = /page\

.thenComposeAsync

(crawlHyperLinks

(page))

/imgNum1\ = /page\

.thenApplyAsync

(countImages(page))



(imgNum, imgNum) ->

Integer::sum)


http://docs.oracle.com/javase/8/docs/api/java/util/concurrent/CompletionStage.html

20


1. Parallel streams



• Async operations can runin parallel in thread pools

See www.nurkiewicz.com/2013/05/java-8-definitive-guide-to.html

/page\ =

supplyAsync

(getStartPage())

/imgNum2\ = /page\

.thenComposeAsync

(crawlHyperLinks

(page))

/imgNum1\ = /page\

.thenApplyAsync

(countImages(page))



(imgNum, imgNum) ->

Integer::sum)


http://www.nurkiewicz.com/2013/05/java-8-definitive-guide-to.html

21


1. Parallel streams



• Async operations can runin parallel in thread pools

Java 8 completable futures & streams can be combined to good effects!!

List of URLs to Download

…

filter(this::nonNull)

collect(toFuture())

map(this::downloadImageAsync)

flatMap(this::applyFiltersAsync)

map(this::checkUrlCachedAsync)


22

• These frameworks often eliminate the use of synchronization or explicit threading when developing parallel apps!

Alleviates many accidental & inherent complexities of parallel programming


23

• Both Java 8 frameworks use the fork-join pool framework by default

See www.oracle.com/technetwork/articles/java/fork-join-422606.html


http://www.oracle.com/technetwork/articles/java/fork-join-422606.html

24

Summary of Example Case Study Apps

25

• SearchStreamGang case study uses regular expressions to find phrases in the complete works of William Shakespeare in parallel

See github.com/douglascraigschmidt/LiveLessons/tree/master/SearchStreamGang


Starting SearchStreamGangTest

PARALLEL_SPLITERATOR executed in 409 msecs

COMPLETABLE_FUTURES_INPUTS executed in 426 msecs

COMPLETABLE_FUTURES_PHASES executed in 427 msecs

PARALLEL_STREAMS executed in 437 msecs

PARALLEL_STREAM_PHASES executed in 440 msecs

RXJAVA_PHASES executed in 485 msecs

PARALLEL_STREAM_INPUTS executed in 802 msecs

RXJAVA_INPUTS executed in 866 msecs

SEQUENTIAL_LOOPS executed in 1638 msecs

SEQUENTIAL_STREAM executed in 1958 msecs

Ending SearchStreamGangTest

https://github.com/douglascraigschmidt/LiveLessons/tree/master/SearchStreamGang

26

• ImageCounter recursively crawls web pages counting # of images in parallel


See github.com/douglascraigschmidt/LiveLessons/tree/master/Java8/ex19

imgs2

imgs3

imgs1

uci.png

robot.png

...ka.png

...

wm.jpg

...

oz.jpg ...vette.

jpg

http://github.com/douglascraigschmidt/LiveLessons/tree/master/Java8/ex19

27See github.com/douglascraigschmidt/LiveLessons/tree/master/ImageStreamGang

Summary of Example Case Study Apps• ImageStreamGang shows how the StreamGang framework can be combined

with Java 8 streams & completable futures to download,filter, store, & display images in parallel

collect(toList())

Parallel Streams

…




ImageStreamGang

https://github.com/douglascraigschmidt/LiveLessons/tree/master/ImageStreamGang

28

End of Introduction to Java 8 Parallelism Frameworks

introduction to java 8 parallelism...

Documents