| commit | 722eaf6eb13efade8e80552b9ef07f611e6ddc1a | [log] [tgz] |
|---|---|---|
| author | arnabp <arnab.phani@tugraz.at> | Mon Jun 01 22:54:02 2020 +0200 |
| committer | Matthias Boehm <mboehm7@gmail.com> | Mon Jun 01 22:54:02 2020 +0200 |
| tree | e3f449b8a6160fac3541c5490563562afa0f3b52 | |
| parent | 8d020fb9f807750b07d08398caa2d433305819b6 [diff] |
[SYSTEMDS-411] Efficient multi-level lineage cache management This patch improves the handling of multiple cache entries pointing to the same data (due to multilevel caching). 1) All the entries with the same values are connected with a linkedlist. Even though they output same data, they have different computation time. 2) Eviction logic marks an entry for deferred spilling/removal if other entries are linked to that. If all the entries in a list are marked for spilling or removal, only then we evict the item. 3) Disk write and read happen only once for all the items connected to a single matrix. This way single read and write restores multiple entries to cache and clears more space respectively. 4) Initial experiments show huge improvements in cache management. Now the cache can store many more entries (this patch fixes duplicate size calculations), need reduced number of disk I/O. These changes overall improve cache hit count. Closes #932.
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.
