Book Image

R High Performance Programming

By : Aloysius Shao Qin Lim
Book Image

R High Performance Programming

By: Aloysius Shao Qin Lim

Overview of this book

This book is for programmers and developers who want to improve the performance of their R programs by making them run faster with large data sets or who are trying to solve a pesky performance problem.

Related Content you might be interested in

Current Title:

Small book image
R High Performance Programming
Aloysius Shao Qin Lim
Jan, 2015 | 176 pages
No titles found
Table of Contents (12 chapters)
Preface
Preface
What this book covers
What you need for this book
Who this book is for
Conventions
Reader feedback
Customer support
Free Chapter
1
1. Understanding R's Performance – Why Are R Programs Sometimes Slow?
1. Understanding R's Performance – Why Are R Programs Sometimes Slow?
Three constraints on computing performance – CPU, RAM, and disk I/O
R is interpreted on the fly
R is single-threaded
R requires all data to be loaded into memory
Algorithm design affects time and space complexity
Summary
2
2. Profiling – Measuring Code's Performance
2. Profiling – Measuring Code's Performance
Measuring total execution time
Profiling the execution time
Profiling memory utilization
Monitoring memory utilization, CPU utilization, and disk I/O using OS tools
Identifying and resolving bottlenecks
Summary
3
3. Simple Tweaks to Make R Run Faster
3. Simple Tweaks to Make R Run Faster
Vectorization
Use of built-in functions
Preallocating memory
Use of simpler data structures
Use of hash tables for frequent lookups on large data
Seeking fast alternative packages in CRAN
Summary
4
4. Using Compiled Code for Greater Speed
4. Using Compiled Code for Greater Speed
Compiling R code before execution
Using compiled languages in R
Summary
5
5. Using GPUs to Run R Even Faster
5. Using GPUs to Run R Even Faster
General purpose computing on GPUs
R and GPUs
Fast statistical modeling in R with gputools
Summary
6
6. Simple Tweaks to Use Less RAM
6. Simple Tweaks to Use Less RAM
Reusing objects without taking up more memory
Removing intermediate data when it is no longer needed
Calculating values on the fly instead of storing them persistently
Swapping active and nonactive data
Summary
7
7. Processing Large Datasets with Limited RAM
7. Processing Large Datasets with Limited RAM
Using memory-efficient data structures
Using memory-mapped files and processing data in chunks
Summary
8
8. Multiplying Performance with Parallel Computing
8. Multiplying Performance with Parallel Computing
Data parallelism versus task parallelism
Implementing data parallel algorithms
Implementing task parallel algorithms
Executing tasks in parallel on a cluster of computers
Shared memory versus distributed memory parallelism
Optimizing parallel performance
Summary
9
9. Offloading Data Processing to Database Systems
9. Offloading Data Processing to Database Systems
Extracting data into R versus processing data in a database
Preprocessing data in a relational database using SQL
Converting R expressions to SQL
Running statistical and machine learning algorithms in a database
Using columnar databases for improved performance
Using array databases for maximum scientific-computing performance
Summary
10
10. R and Big Data
10. R and Big Data
Understanding Hadoop
Setting up Hadoop on Amazon Web Services
Processing large datasets in batches using Hadoop
Summary
11
Index
Index

Summary

In this chapter, we learned how to set up a Hadoop cluster on Amazon Elastic MapReduce, and how to use the RHadoop family of packages in order to analyze data in HDFS using MapReduce. We saw how the performance of the MapReduce task improves dramatically as more servers are added to the Hadoop cluster, but the performance eventually reaches a limit due to Amdahl's law (Chapter 8, Multiplying Performance with Parallel Computing).

Hadoop and its ecosystem of tools is rapidly evolving. Other tools are being actively developed to make Hadoop perform even better. For example, Apache Spark (http://spark.apache.org/) provides Resilient Distributed Datasets (RDDs) that store data in memory across a Hadoop cluster. This allows data to be read from HDFS once and to be used many times in order to dramatically improve the performance of interactive tasks like data exploration and iterative algorithms like gradient descent or k-means clustering. Another example is Apache Storm (http://storm.incubator...

Previous Section
End of Section 2
Next Section