Re: Java HTML Tidy and Hotspot

On Tue, 30 Nov 1999, Andy Quick wrote:

> I have done some benchmarking with Java HTML Tidy
> and Sun's Hotspot VM, and I thought I would share
> the results here.
> 
> I used the program below to parse a number of
> documents by cycling through a list.  In order
> to minimize the effect of runtime library code
> (ie. mostly file and standard output) I simply
> parsed each document without pretty-printing it.
> Also, I set the Errout stream to a CharArrayWriter
> and just threw away the results each time.
> 
> I compared the Sun Hotspot 1.0.1 VM running with
> default settings with the Sun Classic VM with no
> JIT.  The tests were conducted on a clunky Pentium
> Windows 95 machine.
> 
> The results are very impressive, see the table below:
> 
> No. of         Classic      Hotspot     Percent
> Documents      no JIT msec  msec        Improvement
> ---------      -----------  ----------  -----------
>   10             2030        4880       --
>   50            12250       14390       --
>  100            22250       20100       10
>  500           110290       48330       56
> 1000           217010       70030       68

Thanks for the info.

BTW are you updating the Java version to track the improvements
and bugfixes I am making to the C version?


Regards,

-- Dave Raggett <dsr@w3.org> http://www.w3.org/People/Raggett
tel/fax: +44 122 578 3011 (or 2521) +44 385 320 444 (mobile)
World Wide Web Consortium (on assignment from HP Labs)

Received on Wednesday, 1 December 1999 12:06:33 UTC