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

Posted to user@madlib.apache.org by Frank McQuillan <fm...@pivotal.io> on 2016/01/11 19:29:05 UTC

Regarding matrix operations performance

Hi Gautam,

Regarding the JIRA for matrix operations performance
https://issues.apache.org/jira/browse/MADLIB-945

the 8525 x 11 matrix size that you tested is small so we cannot extrapolate
run times from it, due to database start up costs.

As you know, MADlib is designed for large data sets.

Since you have a setup going, can you do runs for:

100,000 rows X 10 columns (i.e., 8 GB for double precision)
1,000,000 rows X 10 columns (80 GB)
10,000,000 rows X 10 columns (800 GB)

and post the results in the JIRA?

Thanks,
Frank