Katana Graph Performance
Katana Graph derives performance advantages from innovations, graph algorithms, storage technologies, and hardware acceleration. The Katana Graph engine has enjoyed a decade of iterative research from founder Keshav Pingali’s group at UT Austin and sets a new standard for performance as the world’s fastest graph processing engine.
Katana Graph scales well on a single machine for the datasets that fit in memory.
Katana Graph seamlessly scales out to a large number of machines in a distributed cluster.
Next Generation Hardware (GPU)
Katana Graph supports single machine multi-GPUs as well as distributed GPU clusters.
Algorithms to analyze global properties or structural patterns of interest in graphs/networks where vertices represent entities of interest and edges represent relationships between these entities. These have applications in a wide range of domains such as chemical engineering, bioinformatics, and social sciences. Some of the applications include financial fraud detection, influencer analysis in social network communities, and optimizing supply distribution chains and logistics.
Algorithms to find the shortest route between two vertices of a graph
- Breadth-First Search
- Single source shortest path
Algorithms to identify vertices of importance a graph
Algorithms to find densely connected communities in a graph
- Connected Components
- Louvain Clustering
- Leiden Clustering
Algorithms to discover graph structural properties
Algorithms for offline training of recommender system
Count/list the number of triangles in a graph
Count/list the complete (all vertices are connected to each other) subgraphs of size K
Count/list the occurrences of a given subgraph in a graph
Count/list the number of occurrences of different subgraphs that are possible with K vertices in a graph
Count/list all the frequently occurring subgraphs with fewer than K edges
Driving the next generation of graph computing
We believe in the power of data. At Katana Graph we are developing technologies to help people and businesses unleash the immense potential of their large scale irregular and unstructured data.