FlashGraph

FlashMatrix

FlashR

FlashX is a collection of big data analytics tools. FlashX performs data analytics in the form of graphs and matrices and utilize solid-state drives (SSDs) to scale to large datasets in a single machine. It has three main components: FlashGraph is a general-purpose graph analysis framework that allows users to write graph algorithms to analyze billion-node graphs in a single machine; FlashMatrix is a matrix computation engine with a small set of generalized matrix operations to express varieties of data mining and machine learning algorithms; FlashR is an extended R programming framework to process datasets at a scale of terabytes in parallel. All code is released under Apache License v2 and is stored at the Github repository.

Performance

Each of FlashX components has lightning-fast speed both in memory and on SSDs. FlashR runs R code with performance comparable to the optimized parallel C code.

Scalability

FlashX components scale beyond memory capacity to graphs with hundreds of billions of edges or matrices of terabytes in a single commodity machine.

Flexibility and ease of programming

FlashX provides simple R programming interface. Parallelization and external-memory data access are completely hidden from users.

Components

FlashGraph

A general-purpose programming framework with a vertex-centric programming interface for large-scale graph analysis. FlashGraph is able to scale to billion-node graphs in a single machine and significantly outperforms state-of-art distributed graph analysis frameworks at this scale.

FlashMatrix

A matrix computation engine that provides a small set of generalized matrix operations to express varieties of data mining and machine learning algorithms. It keeps matrices on SSDs to scale to very large datasets.

FlashR

An extension of the R programming framework to process datasets at a scale of terabytes with the speed of optimized parallel C code. It provides users a familiar R programming environment to express many machine learning algorithms completely in R.