Re: New embedded Java-based controller released

Martin Schoeberl wrote:
> "Ralph.White" <ralph.white@gmail.com> schrieb im Newsbeitrag
> news:c021e1c1.0409080939.c98e4fe@posting.google.com...
> > "Martin Schoeberl" <martin.schoeberl@chello.at> wrote in message
> news:<6zy%c.244062$vG5.188722@news.chello.at>...
> > >
> > > Has someone access to this system? I would be interested how this
JIT
> > > solution on a ARM compares to a 'real' Java processor at about
the
> same
> > > frequency.
> >
> > What would you like to know?
>
> I would like to have some benchmarks running on this board. If you
have
> access to the board I can send you the benchmarks or can put it on my
> website.

I have an evaluation kit with me. If you put the benchmarks on your
site I'll have a look and send the results back.

I am also evaluating other solutions, can I have the results of your
benchmarks for other systems too?

Thanks.

>
> I have an evaluation kit with me. If you put the benchmarks on your
> site I'll have a look and send the results back.
>
Ralph,

here is a link to the benchmarks:
http://www.jopdesign.com/emb_bench.zip

They should be straight-forward to use:

javac jbe/DoAll.java
java jbe.DoAll

runns all benchmarks. Kfl is the application that I have adapted for a
benchmark. If you are interested in what type the application is, the
first paper on JOP (available on the website) describes the application.
Ejip is a tiny TCP/IP stack for embedded Java and the benchmark contains
two UDP server/clients.

Basically all benchmarks are a kind of 'self adjusting' for a great
variation of platfrom performances. Each benchmarks consists of an aspect
that is benchmarked in a loop and an 'overhaed' loop that contains any
overhead from the benchmark that should be subtracted from the result
(this feature is for the micro benchmarks).
The loop count adapts until the benchmark runs for more than a second.

If you need to adapt the benchmark for an embedded device you only have
to change LowLevel.java (see an example for leJOS in the subdircetory).

> I am also evaluating other solutions, can I have the results of your
> benchmarks for other systems too?
>

I will put some results (JOP, JStamp, gcj on small embedded systems...)
on the website.

Martin
--
----------------------------------------------
JOP - a Java Processor core for FPGAs:
http://www.jopdesign.com/