Haskell High Performance Programming

Samuli Thomasson

• Publisher: Packt Publishing
• Year: 2016
• ISBN:
  9781786464217
• Edition: 1
• Topic: Ciencia de la computación
• Subtopic: Desarrollo de software

Boost the performance of your Haskell applications using optimization, concurrency, and parallel programming

About This Book

• Explore the benefits of lazy evaluation, compiler features, and tools and libraries designed for high performance
• Write fast programs at extremely high levels of abstraction
• Work through practical examples that will help you address the challenges of writing efficient code

Who This Book Is For

To get the most out of this book, you need to have a working knowledge of reading and writing basic Haskell. No knowledge of performance, optimization, or concurrency is required.

What You Will Learn

• Program idiomatic Haskell that's also surprisingly efficient
• Improve performance of your code with data parallelism, inlining, and strictness annotations
• Profile your programs to identify space leaks and missed opportunities for optimization
• Find out how to choose the most efficient data and control structures
• Optimize the Glasgow Haskell Compiler and runtime system for specific programs
• See how to smoothly drop to lower abstractions wherever necessary
• Execute programming for the GPU with Accelerate
• Implement programming to easily scale to the cloud with Cloud Haskell

In Detail

Haskell, with its power to optimize the code and its high performance, is a natural candidate for high performance programming. It is especially well suited to stacking abstractions high with a relatively low performance cost. This book addresses the challenges of writing efficient code with lazy evaluation and techniques often used to optimize the performance of Haskell programs.

We open with an in-depth look at the evaluation of Haskell expressions and discuss optimization and benchmarking. You will learn to use parallelism and we'll explore the concept of streaming. We'll demonstrate the benefits of running multithreaded and concurrent applications. Next we'll guide you through various profiling tools that will help you identify performance issues in your program. We'll end our journey by looking at GPGPU, Cloud and Functional Reactive Programming in Haskell. At the very end there is a catalogue of robust library recommendations with code samples.

By the end of the book, you will be able to boost the performance of any app and prepare it to stand up to real-world punishment.

Style and approach

This easy-to-follow guide teaches new practices and techniques to optimize your code, and then moves towards more advanced ways to effectively write efficient Haskell code. Small and simple practical examples will help you test the concepts yourself, and you will be able to easily adapt them for any application.

Haskell High Performance Programming

---

Table of Contents

Haskell High Performance Programming

Credits

About the Author

About the Reviewer

www.PacktPub.com

eBooks, discount offers, and more

Why subscribe?

Preface

What this book covers

What you need for this book

Who this book is for

Conventions

Reader feedback

Customer support

Downloading the example code

Downloading the color images of this book

Errata

Piracy

Questions

1. Identifying Bottlenecks

Meeting lazy evaluation

Writing sum correctly

Weak head normal form

Folding correctly

Memoization and CAFs

Constant applicative form

Recursion and accumulators

The worker/wrapper idiom

Guarded recursion

Accumulator parameters

Inspecting time and space usage

Increasing sharing and minimizing allocation

Compiler code optimizations

Inlining and stream fusion

Polymorphism performance

Partial functions

Summary

2. Choosing the Correct Data Structures

Annotating strictness and unpacking datatype fields

Unbox with UNPACK

Using anonymous tuples

Performance of GADTs and branching

Handling numerical data

Handling binary and textual data

Representing bit arrays

Handling bytes and blobs of bytes

Working with characters and strings

Using the text library

Builders for iterative construction

Builders for strings

Handling sequential data

Using difference lists

Difference list performance

Difference list with the Writer monad

Using zippers

Accessing both ends fast with Seq

Handling tabular data

Using the vector package

Handling sparse data

Using the containers package

Using the unordered-containers package

Ephemeral data structures

Mutable references are slow

Using mutable arrays

Using mutable vectors

Bubble sort with vectors

Working with monads and monad stacks

The list monad and its transformer

Free monads

Working with monad transformers

Speedup via continuation-passing style

Summary

3. Profile and Benchmark to Your Heart's Content

Profiling time and allocations

Setting cost centres manually

Setting cost centres automatically

Installing libraries with profiling

Debugging unexpected crashes with profiler

Heap profiling

Cost centre-based heap profiling

Objects outside the heap

Retainer profiling

Biographical profiling

Benchmarking using the criterion library

Profile and monitor in real time

Monitoring over HTTP with ekg

Summary

4. The Devil's in the Detail

The anatomy of a Haskell project

Useful fields and flags in cabal files

Test suites and benchmarks

Using the stack tool

Multi-package projects

Erroring and handling exceptions

Handling synchronous errors

The exception hierarchy

Handling asynchronous errors

Throw and catch in other monads besides IO

Writing tests for Haskell

Property checks

Unit testing with HUnit

Test frameworks

Trivia at term-level

Coding in GHC PrimOps

Control inlining

Using rewrite rules

Specializing definitions

Phase control

Trivia at type-level

Phantom types

Functional dependencies

Type families and associated types

Useful GHC extensions

Monomorphism Restriction

Extensions for patterns and guards

Strict-by-default Haskell

Summary

5. Parallelize for Performance

Primitive parallelism and the Runtime System

Spark away

Subtle evaluation – pseq

When in doubt, use the force

The Eval monad and strategies

Composing strategies

Fine-tune granularity with chunking and buffering

The Par monad and schedules

spawn for futures and promises

Non-deterministic parallelism with ParIO

Diagnosing parallelism – ThreadScope

Data parallel programming – Repa

Playing with Repa in GHCi

Mapping and delayed arrays

Reduction via folding

Manifest representations

Delayed representation and fusion

Indices, slicing, and extending arrays

Convolution with stencils

Cursored and partitioned arrays

Writing fast Repa code

Additional libraries

Example from image processing

Loading the image from file

Identifying letters with convolution

Extracting strings from an image

Testing and evaluating performance

Summary

6. I/O and Streaming

Reading, writing, and handling resources

Traps of lazy I/O

File handles, buffering, and encoding

Binary I/O

Textual I/O

I/O performance with filesystem objects

Sockets and networking

Acting as a TCP/IP client

Acting as a TCP server (Unix domain sockets)

Raw UDP traffic

Networking above the transport layer

Managing resources with ResourceT

Streaming with side-effects

Choosing a streaming library

Simple streaming using io-streams

Creating input streams

Using combinators and output streams

Handling exceptions and resources in streams

An example of parsing using io-streams and attoparsec

Streaming using pipes

Composing and executing pipes

For loops and category theory in pipes

Handling exceptions in pipes

Strengths and weaknesses of pipes

Streaming using conduits

Handling resources and exceptions in conduits

Resuming conduits

Logging in Haskell

Logging with FastLogger

More abstract loggers

Timed log messages

Monadic logging

Customizing monadic loggers

Summary

7. Concurrency and Performance

Threads and concurrency primitives

Threads and mutable references

Avoid accumulating thunks

Atomic operations with IORefs

MVar

MVars are fair

MVar as a building block

Broadcasting with Chan

Software Transactional Memory

STM example – Bank accounts

Alternative transactions

Exceptions in STM

Runtime System and threads

Masking asynchronous exceptions

Asynchronous processing

Using the Async API

Async example – Timeouts

Composing with Concurrently

Lifting up from I/O

Top-level mutable references

Lifting from a base monad

Lifting base with exception handling

Summary

8. Tweaking the Compiler and Runtime System (GHC)

Using GHC like a pro

Operating GHC

Circular dependencies

Adjusting optimizations and transformations

The state hack

Floating lets in and out

Eliminating common subexpressions

Liberate-case duplicates code

Compiling via the LLVM route

Linking and building shared libraries

Preprocessing Haskell source code

Enforcing type-safety using Safe Haskell

Tuning GHC's Runtime System

Scheduler and green threads

Sparks and spark pool

Bounded threads and affinity

Indefinite blocking...