|
Boost : |
From: Arkadiy Vertleyb (vertleyb_at_[hidden])
Date: 2005-10-17 10:50:37
"Jose" <jmalv04_at_[hidden]> wrote
> Sorry, I forgot to mention that your query was 6 seconds on my x86_64 3Gb
> box
Is this a latest one -- I expect it to take a little longer now... Also I
am curious how long it took to re-run the query after the update.
> I didn't want to compare any results yet because geographical queries
really
> need special
> auxiliary coding, e.g. like building a cities worldmap before actually
> running the queries.
Right, so I am afraid we may fall into the trap of comparing the cleverness
of hand-coded solutions rather than the libraries' capabilities.
> So once I have your updated code I will run both on identical setting.
>
> Calum's is the code author so I am sure he'll post that but I propose the
> solutions are coded separately so that the best ideas come up and then
after
> the benchmark the code is distributed for further analysis, otherwise we
> have some bias in the benchmark. I just wanted to bring up the worldmap
> optimization so that the relational benchmarks can be compared on equal
> terms
>
> I will email you directly some of my own optimization ideas as I have not
> yet looked in detail at calum's code
The reason I wanted to take a look at it is that I want to see if it looks
more like a library-based solution or a hand-coded solution. Also, I want
to try more relational-based optimizations before starting to apply other
tricks.
But, again, could you see how long the re-run (after update) takes?
Thanks,
Arkadiy
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk