[PATCH] speed up tokenize_command

[Marc]

You¹re compiling with ­02 and all the changes you made look like the kind of
optimizations that ­03 will do for you automatically only better (since ­O3
will do the right thing across platforms).  Can you verify that you¹re
actually measure a speedup with ­O3?

On 4/12/07 4:57 PM, "Paolo Borelli" <paolo.borelli at gmail.com> wrote:

> Hi, thanks for integrating the previous patches so fast!
> 
> It made me want to stare a bit at the code to look for any other little
> improvements to make. I stumbled on tokenize_command() and for fun I
> tried to write a faster version (no idea if it is a real bottleneck).
> 
> Attached there is my version and the small bench I used: on my machine
> the function is about 20% faster. The nice thing is that the function is
> even shorter and cleaner :)
> 
> I tested it a bit to see if my function handled all the cases in the
> same way as the old one and it seems ok, but keep in mind that it has
> not been tested in production...
> 
> 
> ciao
>  Paolo


-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.danga.com/pipermail/memcached/attachments/20070416/2e516a63/attachment.htm