Re: Speed comparison between vanilla pyyaml and pyyaml+libyaml

On Tue, Apr 21, 2009 at 9:52 PM,  <[email protected]> wrote:
> I ran across the following posts a little while ago:
>
> I'm a heavy yaml consumer, so I thought I'd experiment with vanilla YAML
> speed runs of the same benchmark, and then with pyyaml+libyaml, with cython
> installed. What bothers me is that there's a negligible difference in speed,
> even with libyaml installed, and pyyaml using the cyaml extension:
>
> Here's the relevant numbers:
>
> Vanilla PyYAML:
>
> ser_yaml (6.320s) 623436 bytes
> ser_yaml_compressed (6.410s) 291050 bytes
> serde_yaml (18.550s)
>
> PyYAML + libyaml:
>
> ser_yaml (6.150s) 623488 bytes
> ser_yaml_compressed (6.240s) 291175 bytes
> serde_yaml (18.950s)
>
> On the serde_yaml test, I've consistently gotten worse speeds than plain old
> pyyaml. I'm wondering if you guys have any thoughts/ideas on this. I've
> posted the modified benchmark code to:
>
> http://jessenoller.com/code/test_yaml.py
>
> FWIW, this is on a dual core OS/X box. In the morning, I'll try running it
> on an available Linux machine just for sanity's sake
>
> jesse

Mea culpa, it's later here, and I forgot to include the links I meant
to reference:

http://www.bouncybouncy.net/ramblings/posts/thrift_and_protocol_buffers/
http://www.bouncybouncy.net/ramblings/posts/more_on_json_vs_thrift_and_protocol_buffers/
http://www.bouncybouncy.net/ramblings/posts/json_vs_thrift_and_protocol_buffers_round_2/

-jesse

------------------------------------------------------------------------------
Stay on top of everything new and different, both inside and 
around Java (TM) technology - register by April 22, and save
$200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco.
300 plus technical and hands-on sessions. Register today. 
Use priority code J9JMT32. http://p.sf.net/sfu/p