[SYSTEMML-2196] Multi-threaded matrix blocking on RDD parallelize

This patch improves the performance for CP to Spark data exchange by
parallelizing the matrix blocking on RDD parallelize. For a scenario of
10 x 800MB matrix parallelization (incl blocking),this patch improved
performance from 8.1s to 3.8s.
1 file changed
tree: 14966365a00cb2bbad5817c74f0d27ce11b32020
  1. bin/
  2. conf/
  3. dev/
  4. docs/
  5. projects/
  6. samples/
  7. scripts/
  8. src/
  9. .gitattributes
  10. .gitignore
  11. LICENSE
  12. NOTICE
  13. pom.xml
  14. README.md
README.md

SystemML

Documentation: SystemML Documentation
Mailing List: Dev Mailing List
Build Status: Build Status
Issue Tracker: JIRA
Download: Download SystemML

SystemML is now an Apache Top Level Project! Please see the Apache SystemML website for more information.

SystemML is a flexible, scalable machine learning system. SystemML's distinguishing characteristics are:

  1. Algorithm customizability via R-like and Python-like languages.
  2. Multiple execution modes, including Spark MLContext API, Spark Batch, Hadoop Batch, Standalone, and JMLC.
  3. Automatic optimization based on data and cluster characteristics to ensure both efficiency and scalability.

The latest version of SystemML supports: Java 8+, Scala 2.11+, Python 2.7/3.5+, Hadoop 2.6+, and Spark 2.1+.

Algorithm Customizability

ML algorithms in SystemML are specified in a high-level, declarative machine learning (DML) language. Algorithms can be expressed in either an R-like syntax or a Python-like syntax. DML includes linear algebra primitives, statistical functions, and additional constructs.

This high-level language significantly increases the productivity of data scientists as it provides (1) full flexibility in expressing custom analytics and (2) data independence from the underlying input formats and physical data representations.

Multiple Execution Modes

SystemML computations can be executed in a variety of different modes. To begin with, SystemML can be operated in Standalone mode on a single machine, allowing data scientists to develop algorithms locally without need of a distributed cluster. In order to scale up, algorithms can also be distributed across a cluster using Spark or Hadoop. This flexibility allows the utilization of an organization's existing resources and expertise. In addition, SystemML features a Spark MLContext API that allows for programmatic interaction via Scala, Python, and Java. SystemML also features an embedded API for scoring models.

Automatic Optimization

Algorithms specified in DML are dynamically compiled and optimized based on data and cluster characteristics using rule-based and cost-based optimization techniques. The optimizer automatically generates hybrid runtime execution plans ranging from in-memory, single-node execution, to distributed computations on Spark or Hadoop. This ensures both efficiency and scalability. Automatic optimization reduces or eliminates the need to hand-tune distributed runtime execution plans and system configurations.

ML Algorithms

SystemML features a suite of production-level examples that can be grouped into six broad categories: Descriptive Statistics, Classification, Clustering, Regression, Matrix Factorization, and Survival Analysis. Detailed descriptions of these algorithms can be found in the SystemML Algorithms Reference. The goal of these provided algorithms is to serve as production-level examples that can modified or used as inspiration for a new custom algorithm.

Download & Setup

Before you get started on SystemML, make sure that your environment is set up and ready to go.

  1. If you’re on OS X, we recommend installing Homebrew if you haven’t already. For Linux users, the Linuxbrew project is equivalent.

OS X:

/usr/bin/ruby -e "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install)"

Linux:

ruby -e "$(curl -fsSL https://raw.githubusercontent.com/Linuxbrew/install/master/install)"
  1. Install Java (need Java 8).
brew tap caskroom/cask
brew install Caskroom/cask/java
  1. Install Spark 2.1.
brew tap homebrew/versions
brew install apache-spark21
  1. Download SystemML.

Go to the SystemML Downloads page, download systemml-1.0.0-bin.zip (should be 2nd), and unzip it to a location of your choice.

The next step is optional, but it will make your life a lot easier.

  1. [OPTIONAL] Set SYSTEMML_HOME in your bash profile. Add the following to ~/.bash_profile, replacing path/to/ with the location of the download in step 4.
export SYSTEMML_HOME=path/to/systemml-1.0.0

Make sure to open a new tab in terminal so that you make sure the changes have been made.

  1. [OPTIONAL] Install Python or Python 3 (to follow along with our Jupyter notebook examples).

Python 2:

brew install python
pip install jupyter matplotlib numpy

Python 3:

brew install python3
pip3 install jupyter matplotlib numpy

Congrats! You can now use SystemML!

Next Steps!

To get started, please consult the SystemML Documentation. We recommend using the Spark MLContext API to run SystemML from Scala or Python using spark-shell, pyspark, or spark-submit.