[Xastir] Perfromance Comparison

[Dan Brown]

On Wed, 2 Feb 2005, Ryan Tourge wrote:

> I have Xastir 1.4.2 running on two computer here. The first is a Sony
> Vaio FXA63 with an AMD Athlon XP 1.4 GHz and 256MB RAM. The PC is home
> built with an AMD Duron 800MHz processor and 320MB RAM.
> 
> Both installations are using proMEPIS and a compiled Xastir is
> configured the same way for both (including map caching).
> 
> On the laptop maps load much faster then the desktop.  Should I just
> assume that it is due to the laptop having more "ponies under the
> hood"? The only reason I ask is the processor activity doesn't show
> that much of difference between the two machines. Or do you think
> there is somethign else I should be looking at? 

There are several possible places where you could see timing differences 
between the two machines, and you'll need to go through a process of 
elimination to figure out which is affecting you. Since the tiger maps 
come from a remote source, there's a potential communications delay and 
there are several places where things could be proxied or cached - not 
just locally! 

The first step would be to make sure you have all the same options clicked 
for the tiger maps - and that you're looking at the same view(s) on both 
systems, etc. One of the biggest timing differences I've seen appears to
come from number of different things that need to be plotted (streets, 
rivers, etc) and, of course, that time will then be multiplied by the 
amount of area you're looking at. A Zoom 1 view takes lots less time to 
plot than a zoom 4096. 

Anyhow - after that, turn on debug to level 512 and look for the 
string:

"Query took ### seconds" 

If the queries come back in the same number of seconds, then you're down 
to trying to figure out differences between the local machines (graphics 
card maybe? Cache on the CPU? I dunno). If the seconds are wildly off, then
you're looking at some difference at tiger.census.gov or someplace in between. 

Does the map loading difference occur always on the same machine? or Does 
it matter which one you do the query from first? That might imply some caching or 
proxy in the Tiger.census system or on the net some place. 

>Once the maps are cached it works fine.

Music to my ears ;-) 

--
Dan Brown 
brown at brauhausdc.org