Hi all,
Rendering speed question for you:
I'm trying to display 3 categories of road labels (major, minor and local) from 
a US-wide street dataset.  In using nik2img.py, I'm able to get reasonable 
speeds (i.e. 7 seconds for 1024x1024 image) at all zoom levels when only major 
and minor roads labels are shown.  But when adding in the local road labels 
which are only shown at zooms 17 and 18, rendering time balloons to over 4 
minutes per 1024x1024 tile.

I would expect a slowdown given the greater number of local road records in my 
database (4.2 million vs 74k and 222k major and minor road records 
respectively), but I'm able to render full Planet OSM data in a similarly 
styled map with full road labels at z17 and 18 in as little as 0.39 seconds on 
the same machine.  Also, rendering local roads themselves is relatively fast; 
it's only when their labels are added that things slow down measurably.

B-Tree and GiST indexes have been built.  I've posted my XML below for 
reference.
Thanks in advance for your help,
Mark





<Rule>
        <Filter>([mtfcc] = 'S1400' or [mtfcc] = 'C3062' or [mtfcc] = 'C3066' or 
[mtfcc] = 'C3067') and [func_class] = 5</Filter>
        &maxscale_zoom17;
        &minscale_zoom17;
        <TextSymbolizer name="[name]" fontset-name="helvetica-medext" size="9" 
fill="#5e5e5e" halo-fill="rgba(255,255,255,1)" halo-radius="1" placement="line" 
max-char-angle-delta="20" spacing="300" minimum-distance="200" 
avoid-edges="true"/>
</Rule>


<Layer name="surface street text" status="on" srs="&srs900913;">
        <StyleName>surface street text</StyleName>
        <Datasource>
                <Parameter name="table">(select 
the_geom,mtfcc,name,func_class,pk_lot_rd,route_name,length_mi from 
streetsdisplay where "name" is not null and not "pk_lot_rd" in (1) and not 
"route_name" in (1)) as text</Parameter>
                &datasource-settings;
        </Datasource>
</Layer>

_______________________________________________
Mapnik-users mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/mapnik-users

Reply via email to