Mikal,
 
Can you send me the change you made and the indexes you added.  When adding
some more data, I realized I had hardcoded an index for our local state (MA)
and I know without that index that that particular query does run pretty
slow.  So just wondering if its along the same lines.
 
I've also fixed I think all the issues with running the loader on Unix/Linux
-- well at least I was able to get it to run on my CentOS.  Thanks to all
who contributed input to that.  I took bits and pieces from many people's
comments but couldn't apply a full diff from anyones since I had already
changed the code too much to safely apply any of those patches.
 
How many states do you have loaded BTW?  I just have CA loaded on my CentOS
-- which is an 8GB/8 core cloud server.  I'm getting around 38ms - 450ms per
test, but I have yet to load the other states.
It also seems to cache very well so that if I geocode an address on the same
street (not necessarily same address), the first call might take 450ms and
the second 38ms.  I suspect this might be because I also marked a good chunk
of the functions STABLE or IMMUTABLE.
 
Thanks,
Regina
http://www.postgis.us
 


  _____  

From: [email protected]
[mailto:[email protected]] On Behalf Of Mikal
Laster
Sent: Friday, May 27, 2011 8:18 AM
To: [email protected]
Subject: [postgis-users] Fw: re: Geocoder (from extras)




--- On Fri, 5/27/11, Mikal Laster <[email protected]> wrote:




From: Mikal Laster <[email protected]>
Subject: re:[postgis-users] Geocoder (from extras)
To: [email protected]
Date: Friday, May 27, 2011, 7:22 AM



in response to
http://postgis.refractions.net/pipermail/postgis-users/2011-May/029566.html.
After creating some indexes and rewriting geocode_address. I was able to get
geocode to run in 483-523 ms for "5775 Perimeter Dr Dublin, Ohio". This used
to take 1700-2000 ms for me. I'm removing the main inner qui
        

_______________________________________________
postgis-users mailing list
[email protected]
http://postgis.refractions.net/mailman/listinfo/postgis-users

Reply via email to