| commit | 75648fe8a3817a4971480b09cce3ae0d694c7d06 | [log] [tgz] |
|---|---|---|
| author | Matthias Boehm <mboehm7@gmail.com> | Tue May 26 20:15:24 2020 +0200 |
| committer | Matthias Boehm <mboehm7@gmail.com> | Tue May 26 20:15:24 2020 +0200 |
| tree | a0a222dfa528eaa9803d21603ad1c695ec23e5e8 | |
| parent | b66a3c006ce6a3a888653e2d1accec479cc756fd [diff] |
[SYSTEMDS-209] Performance sparse matrix-colvector cell-wise multiply While working on the new builtin function for connected components and ultra-sparse graphs, we found that 'rowMaxs(G * t(c))' performed orders of magnitude better than the semantically equivalent 't(colMaxs(G * c))'. The reason was a missing handling of strict sparse-safe operations for matrix-colvector operations, while this was already handled for matrix-rowvector operations. In detail, we performed unnecessary operations in the number of cells instead of in the number of non-zeros leading to worse asymptotic behavior. With the simple fix of this patch, now we have very similar performance. For example, on a scenario of performing 100 times G*c where X is a 10Kx10K, sparsity=0.0001 matrix, total execution time (for 100 operations) improved from 4.2s to 167ms.
Overview: SystemDS is a versatile system for the end-to-end data science lifecycle from data integration, cleaning, and feature engineering, over efficient, local and distributed ML model training, to deployment and serving. To this end, we aim to provide a stack of declarative languages with R-like syntax for (1) the different tasks of the data-science lifecycle, and (2) users with different expertise. These high-level scripts are compiled into hybrid execution plans of local, in-memory CPU and GPU operations, as well as distributed operations on Apache Spark. In contrast to existing systems - that either provide homogeneous tensors or 2D Datasets - and in order to serve the entire data science lifecycle, the underlying data model are DataTensors, i.e., tensors (multi-dimensional arrays) whose first dimension may have a heterogeneous and nested schema.
Quick Start Install, Quick Start and Hello World
Documentation: SystemDS Documentation
Python Documentation Python SystemDS Documentation
Status and Build: SystemDS is still in pre-alpha status. The original code base was forked from Apache SystemML 1.2 in September 2018. We will continue to support linear algebra programs over matrices, while replacing the underlying data model and compiler, as well as substantially extending the supported functionalities. Until the first release, you can build your own snapshot via Apache Maven: mvn clean package -P distribution.
