You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@systemml.apache.org by "Matthias Boehm (JIRA)" <ji...@apache.org> on 2016/10/05 00:37:20 UTC

[jira] [Closed] (SYSTEMML-471) COO sparse matrix block

     [ https://issues.apache.org/jira/browse/SYSTEMML-471?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Matthias Boehm closed SYSTEMML-471.
-----------------------------------

> COO sparse matrix block
> -----------------------
>
>                 Key: SYSTEMML-471
>                 URL: https://issues.apache.org/jira/browse/SYSTEMML-471
>             Project: SystemML
>          Issue Type: Task
>            Reporter: Matthias Boehm
>            Assignee: Matthias Boehm
>             Fix For: SystemML 0.10
>
>
> Given the new sparse matrix block abstraction defined in [SYSTEMML-378], it would be nice to support the COO sparse matrix format for ultra-sparse matrices as well. CSR is memory-efficient for sparse matrices but incurs unnecessary overhead for ultra-sparse matrices. For example, consider a 1k x 1k matrix with a sparsity of (10^-5), i.e, 10 non-zero values. CSR would require roughly 1k*4B+10*12B = 4120B, while COO would require only 10*16B = 160B.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)