BID Data Suite is a collection of hardware, software and design patterns that enable fast, large-scale data mining at very low cost.
The elements of the suite are:
a hardware design pattern that balances storage, CPU and GPU acceleration for typical data mining workloads.
an interactive matrix library that integrates CPU and GPU acceleration and novel computational kernels.
a machine learning system that includes very effcient model optimizers and mixing strategies.
a communication strategy that hides the latency of frequent model updates needed by fast optimizers for clusters.
to improve performance of iterative update algorithms.
In the benchmark section, we present several benchmark problems to show how the above elements combine to yield multiple orders-of-magnitude improvements for each problem.