[plt-scheme] Load times

"mzc --extension" compiles to Native Code via C; "mzc -k" compiles to Bytecode. For PLT, Bytecode is faster than Native Code.


Chongkai

Corey Sweeney wrote:
> Interesting.  I was already compiling things, but your way seems to
> load about 30% faster then "mzc --extension --auto-dir" which I was
> doing.
>
> Corey
>
> On Thu, Sep 18, 2008 at 2:57 PM, Noel Welsh <noelwelsh at gmail.com> wrote:
>   
>> mzc -k your-main-file.ss
>>
>> You will see a dramatic difference in load time, as MzScheme can read
>> bytecode MUCH faster than it can read and compile Scheme source.
>>
>> N.
>>
>> On Thu, Sep 18, 2008 at 8:42 PM, Corey Sweeney <corey.sweeney at gmail.com> wrote:
>>     
>>> I've been experiencing worse and worse load times, as I write more and
>>> more libraries.  Many of my programs are supposed to run for a second,
>>> then quit, so the load times are significant.  I've been running some
>>> compilation and modularization expirments, and havn't had much luck.
>>> If I understand correctly, compiling with "mzc" does not make a
>>> "static" binary, so when the code is run, all the functions that in
>>> the libraries that aren't used by the application are still loaded.
>>>
>>> Are there any recomendations for reducing load times?  Is there any
>>> way to strip out unused functions?
>>>
>>> Corey
>>>       
>
>
>
>