Matthias Boehm updated SYSTEMML-2173:
    Issue Type: Sub-task  (was: Bug)
        Parent: SYSTEMML-2061

> Upper bound nnz propagation over spark instructions
> ---------------------------------------------------
>                 Key: SYSTEMML-2173
>                 URL: https://issues.apache.org/jira/browse/SYSTEMML-2173
>             Project: SystemML
>          Issue Type: Sub-task
>            Reporter: Matthias Boehm
>            Priority: Major
> Due to lazy evaluation, many spark instructions propagate nnz only if they 
> can be exactly inferred from the input meta data. Especially for ultra-sparse 
> matrices this poses a challenge because intermediates quickly reach a state 
> where they're assumed to be dense. However, many operations such as right 
> indexing allow to upper bound the nnz with the nnz in the input, which can be 
> propagated and exploited in subsequent instructions.

This message was sent by Atlassian JIRA

Reply via email to