Re: [Gzz] Benchmarks

[Matti Katila]

Well,


DEMO=coords_benchmarks working with 100 iterations per timeRender,
results are here:
  http://people.cc.jyu.fi/~majukati/text_benchmarks

According to data:

On Mon, 17 Mar 2003, Tuomas Lukka wrote:
> Also, if text is the problem, I'll need to know the scaling:
> how does it scale with
> 
> 1) increasing number of letters in the text
> 2) increasing number of texts of same size
> 3) increasing depth of nested coordinate systems.
> 
> Because if 1) scales linearly, I'll know that transformations
> are too slow.

It scales linearly.

> If 1) is almost constant but 2) scales badly,
> I'll know that the text init code is slow.

humm...

>  If 3) scales badly, I'll know that the problem is in the tr() method.

I think it does not scale badly.


   -Matti