Hi, Jong,

The index construction time is proportional the number of rows to be
indexed even in the best case.  What I see if that the time basically
doubles as you increase the number of rows.  In this sense, FastBit is
doing OK.

With that said, if you can tell us what type of index you are
building, we might be able to give you more precise recommendations.

John

PS: The available indexing options from FastBit is described at
<http://lbl.gov/~kwu/fastbit/doc/indexSpec.html>.  Near the end of
that document, there is a set of recommendations.



On 3/28/13 10:07 AM, Jong Choi wrote:
> Hi John,
> 
> I have just got a performance result measured in SITH, our local
> cluster, by using a single core for indexing, querying, and sending
> data to LBNL (attached). It looks like indexing takes a lot. Is there
> any way to do parallel in-memory indexing? I saw some examples using
> multi-threading but am having a hard time to do in a in-memory way
> from the example you sent me before.
> 
> Thank you for your help in advance.
> 
> Thanks,
> Jong
> 
_______________________________________________
FastBit-users mailing list
[email protected]
https://hpcrdm.lbl.gov/cgi-bin/mailman/listinfo/fastbit-users

Reply via email to