Concurrency Theory vs Concurrent Languages Silvia Crafa Universita - PowerPoint PPT Presentation

Concurrency Theory vs Concurrent Languages Silvia Crafa � Universita’ di Padova � Bertinoro, OPCT 2014

Bisimulation inside Concurrency Theory vs Concurrent Languages Silvia Crafa � Universita’ di Padova � Bertinoro, OPCT 2014

The Quest for good Abstractions Fortran � Pascal C � C++ � JavaScript Go Java C# Lisp � Ruby � ML Haskell Scala X10 PHP Cobol Python ✤ When a language has been invented VS when became popular ? � ✤ Why has been invented VS why became popular?

The Quest for good Abstractions Add structure � to the code Fortran � Pascal C � C++ � JavaScript Go Java C# Lisp � Ruby � ML Haskell Scala X10 PHP Cobol Python ✤ When a language has been invented VS when became popular ? � ✤ Why has been invented VS why became popular?

The Quest for good Abstractions OOP to handle � industrial complex � software systems � Encapsulation/Modularity Add structure � Interfaces/Code Reuse to the code Fortran � Pascal C � C++ � JavaScript Go Java C# Lisp � Ruby � ML Haskell Scala X10 PHP Cobol Python ✤ When a language has been invented VS when became popular ? � ✤ Why has been invented VS why became popular?

The Quest for good Abstractions OOP to handle � industrial complex � software systems � Encapsulation/Modularity Add structure � Interfaces/Code Reuse to the code Fortran � Pascal C � C++ � JavaScript Go Java C# Lisp � Ruby � ML Haskell Scala X10 PHP Cobol Python INTERNET ✤ When a language has been invented VS when became popular ? � ✤ Why has been invented VS why became popular?

The Quest for good Abstractions less Efficiency � OOP to handle � industrial complex � more Portability � software systems � Security (Types) � Encapsulation/Modularity Add structure � GUIs and IDEs Interfaces/Code Reuse to the code Fortran � Pascal C � C++ � JavaScript Go Java C# Lisp � Ruby � ML Haskell Scala X10 PHP Cobol Python INTERNET ✤ When a language has been invented VS when became popular ? � ✤ Why has been invented VS why became popular?

The Quest for good Abstractions less Efficiency � OOP to handle � industrial complex � more Portability � software systems � Security (Types) � Productivity � Encapsulation/Modularity Add structure � Types are burdensome GUIs and IDEs Interfaces/Code Reuse to the code Fortran � Pascal C � C++ � JavaScript Go Java C# Lisp � Ruby � ML Haskell Scala X10 PHP Cobol Python INTERNET ✤ When a language has been invented VS when became popular ? � ✤ Why has been invented VS why became popular?

The Quest for good Abstractions less Efficiency � OOP to handle � industrial complex � more Portability � software systems � Security (Types) � Productivity � Encapsulation/Modularity Add structure � Types are burdensome GUIs and IDEs Interfaces/Code Reuse to the code Fortran � Pascal C � C++ � JavaScript Go Java C# Lisp � Ruby � ML Haskell Scala X10 PHP Cobol Python INTERNET CONCURRENCY ✤ When a language has been invented VS when became popular ? � ✤ Why has been invented VS why became popular?

The Quest for good Abstractions Fortran � Pascal C � C++ � JavaScript Go Java C# Lisp � Ruby � ML Haskell Scala X10 PHP Cobol Python INTERNET CONCURRENCY ✤ Changes need a catalyser

The Quest for good Abstractions Fortran � Pascal C � C++ � JavaScript Go Java C# Lisp � Ruby � ML Haskell Scala X10 PHP Cobol Python INTERNET CONCURRENCY ✤ Changes need a catalyser ✤ new hardware can only be parallel � ✤ new software must be concurrent

The Quest for good Abstractions Fortran � Pascal C � C++ � JavaScript Go Java C# Lisp � Ruby � ML Haskell Scala X10 PHP Cobol Python INTERNET CONCURRENCY Popular ✤ Changes need a catalyser Parallel Programming ✤ new hardware can only be parallel � Grand Challenge ✤ new software must be concurrent

How hard is Concurrent Programming? ✤ (correct) concurrent programming is difficult � Intrinsic reasons � nondeterminism ✤ Adding concurrency to sequential code is even harder Accidental reasons � improper � programming model

How hard is Concurrent Programming? ✤ (correct) concurrent programming is difficult � Intrinsic reasons � nondeterminism ✤ Adding concurrency to sequential code is even harder Accidental reasons � improper � programming model Think concurrently � Translate into � (Concurrent Algorithm) a concurrent code

How hard is Concurrent Programming? ✤ (correct) concurrent programming is difficult � Intrinsic reasons � nondeterminism ✤ Adding concurrency to sequential code is even harder Accidental reasons � improper � programming model Think concurrently � Translate into � (Concurrent Algorithm) a concurrent code DESIGN of � concurrent language

How hard is Concurrent Programming? ✤ (correct) concurrent programming is difficult � Intrinsic reasons � nondeterminism ✤ Adding concurrency to sequential code is even harder Accidental reasons � improper � programming model Think concurrently � Translate into � (Concurrent Algorithm) a concurrent code High-level DESIGN of � concurrent language Concurrency Abstraction

The Quest for good Abstractions Expressiveness � Easy to think � Performance Easy to reason about

The Quest for good Abstractions Expressiveness � Easy to think � Performance Easy to reason about ✤ OOP � encapsulation � ✤ memory management � ✤ multiple inheritance ✤

The Quest for good Abstractions Expressiveness � Easy to think � Performance Easy to reason about ✤ OOP � encapsulation � ✤ memory management � ✤ multiple inheritance ✤ C++ —> Java —> Scala

The Quest for good Abstractions Expressiveness � Easy to think � Performance Easy to reason about ✤ OOP � encapsulation � ✤ ✤ Types � memory management � ✤ multiple inheritance documentation vs verbosity ✤ ✤ C++ —> Java —> Ruby —>Scala C++ —> Java —> Scala

The Quest for good Abstractions Expressiveness � Easy to think � Performance Easy to reason about ✤ OOP � encapsulation � ✤ ✤ Types � memory management � ✤ multiple inheritance documentation vs verbosity ✤ ✤ C++ —> Java —> Ruby —>Scala C++ —> Java —> Scala ✤ Functional Programming � composing and passing behaviours � ✤ sometimes imperative style is easier to reason about ✤ C#—> Scala C++11, Java8

The Quest for good Abstractions Expressiveness � Easy to think � Performance Easy to reason about ✤ OOP which abstractions � interoperate � ✤ Types productively? ✤ Functional Programming

The Quest for good Abstractions Concurrency Abstractions? � Many Concurreny Models…

The Quest for good Abstractions Concurrency Abstractions? � Many Concurreny Models… ✤ Shared Memory Model and “Java Threads”

The Quest for good Abstractions Concurrency Abstractions? � Many Concurreny Models… ✤ Shared Memory Model and “Java Threads” STM Java X10 synchronized(lock) � atomic {…} � async{} � lock.wait() � when(cond){…} finish{} lock.notify()

The Quest for good Abstractions Concurrency Abstractions? � Many Concurreny Models… ✤ Shared Memory Model and “Java Threads” STM Java X10 synchronized(lock) � atomic {…} � async{} � lock.wait() � when(cond){…} finish{} lock.notify() Scalability! logical threads distinguished from executors ✤ (activities/tasks) (pool of thread workers)

The Quest for good Abstractions Concurrency Abstractions? � Many Concurreny Models… ✤ Shared Memory Model and “Java Threads” STM Java X10 synchronized(lock) � atomic {…} � async{} � lock.wait() � when(cond){…} finish{} lock.notify() Lightweight threads in the program � new Thread().start() � Pool of Executors in the runtime JVM thread Scalability! logical threads distinguished from executors ✤ (activities/tasks) (pool of thread workers)

Many Concurrency Models ✤ Shared Memory � is very natural for “centralised algorithms” ✤ and components operating on shared data � is error-prone when the sole purpose of SM ✤ is thread communication ✤ Message Passing Model � It is the message that carries the state! � ✤ C hannel based : Google’s GO � ✤ Actor Model : Erlang, Scala. It fits well ✤ both OOP and FP � ✤ GPU Concurrency Model � Sessions ✤ Massive data parallelism � ✤ integration with high-level concurrent language ✤ (X10, Nova, Scala heterogeneous compiler)

Many Concurrency Models ✤ Shared Memory which abstractions � interoperate � productively? ✤ Message Passing Model ✤ GPU Concurrency Model

The Quest for good Abstractions Fortran � Pascal C � C++ � JavaScript Go Java C# Lisp � Ruby � ML Haskell Scala X10 PHP Cobol Python INTERNET CONCURRENCY ✤ New catalyser :

The Quest for good Abstractions Fortran � Pascal C � C++ � JavaScript Go Java C# Lisp � Ruby � ML Haskell Scala X10 PHP Cobol Python INTERNET CONCURRENCY DISTRIBUTION ✤ New catalyser : ✤ multicore —> concurrent programming � ✤ cloud computing —> distributed programming