[datatable-help] Memory usage of data.table chaining

[Arunkumar Srinivasan]

It depends. Keys are useful if you’ve to set it once, and use it for repeated subsets, or you’ve really huge data, where keeping data sorted in memory could improve speed tremendous due to cache efficiency.
But auto indexing would be the way to go wherever applicable. We should be expanding it when we find time next.

-- 
Arun

On 21 Feb 2015 at 15:02:25, Mick Cooney (mickcooney at gmail.com) wrote:

I generally don't think of using the key, is it worth setting  
temporary keys for stuff like that?  

I would have thought that if you are doing a select on different  
columns (thus meaning the keys would need to be recreated) that the  
speed up from the key-based select would be negated by the cost of  
resetting keys?  

It's definitely something I should probably consider doing more.  

--  
Mick Cooney  
mickcooney at gmail.com  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/datatable-help/attachments/20150224/5a716013/attachment.html>