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

Posted to dev@sdap.apache.org by "Frank Greguska (JIRA)" <ji...@apache.org> on 2018/03/02 01:43:00 UTC

[jira] [Commented] (SDAP-16) Prototype multidimensional tiles

    [ https://issues.apache.org/jira/browse/SDAP-16?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16382986#comment-16382986 ] 

Frank Greguska commented on SDAP-16:
------------------------------------

Was able to implement some basic xarray usage and time it against the original using local data.

 

Non xarray:

735 ms ± 53.4 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)

 

With xarray

980 ms ± 97.9 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)

> Prototype multidimensional tiles
> --------------------------------
>
>                 Key: SDAP-16
>                 URL: https://issues.apache.org/jira/browse/SDAP-16
>             Project: Apache Science Data Analytics Platform
>          Issue Type: Improvement
>          Components: nexus
>            Reporter: Frank Greguska
>            Assignee: Frank Greguska
>            Priority: Major
>
> The current representation of a tile in NEXUS is a flat 2-D grid indexed by latitude and longitude. This structure is proving to be insufficient for the heterogenous data NEXUS is required to handle.
> We need to experiment with n-dimensional tiles (in reality, 4 dimensions is probably the maximum: latitude, longitude, time, and depth/altitude) to see if we can improve functionality and performance of the system.
> This issue will track changes required to the system in order to handle n-dimensional tiles.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)