Hi, Jan,

The source code of FastBit should be thread-safe, which means that you 
should be able to run multiple query processing functions at the same 
time.  You can use multiple indexes and multiple data partitions at 
the same time too.

In most case, the performance of FastBit is I/O dominated -- reading 
the data files or the index files takes the longest amount of time; 
parallelizing query answering does not reduce the query response time 
in such situation.  If you plan to load the data files and the index 
files into memory, then parallelizing the query processing would be 
helpful.

John


On 6/14/2010 10:35 AM, Krivanek, Jan wrote:
> I saw one similar question regarding parallelization on cluster, so I
> was wandering about parallelization on single PC with more processors
> (cores) – Does FastBit perform such a parallelization for some tasks -
> e.g. when searching multiple indexes or querying multiple partitions.
>
> I quickly inspected mensa select functions which operates on multiple
> partitions and found that those are performed sequentially. Is there
> actually any implementation that parallelize queries, that I should use,
> or the parallelization should be fully handled on client side? If on
> client side, can you recommend an efficient way how to accomplish this
> (I was thinking about querying partitions separately, dumping the data
> in one in-memory partition/table and then perform additional operations
> like grouping, ordering etc., but I am not sure if this would help).
>
> Thank you
>
> Jan Krivanek
>
>
>
> _______________________________________________
> FastBit-users mailing list
> [email protected]
> https://hpcrdm.lbl.gov/cgi-bin/mailman/listinfo/fastbit-users
_______________________________________________
FastBit-users mailing list
[email protected]
https://hpcrdm.lbl.gov/cgi-bin/mailman/listinfo/fastbit-users

Reply via email to