[kaffe] avr32 port

[Nevo]

hi,Guennadi:

Guennadi Liakhovetski wrote:
> On Thu, 8 May 2008, Nevo wrote:
>
>   
>> Guennadi Liakhovetski wrote:
>>     
>>> On Thu, 8 May 2008, Nevo wrote:
>>>   
>>>       
>>>>    There has been a crazy idea for me to  embed kaffe VM or phoneme CDC vm
>>>> to
>>>> Linux kernel as a separate module to speed up vm execution,so vm can
>>>> directly
>>>> access hardware. But you know,that should break up the role of JVM as a
>>>> sandbox...... and we chould not prove the feasibility yet...
>>>>     
>>>>         
>>> Yeah, don't think the kernel crowd would be particularly inspired by this
>>> idea:-)
>>>   
>>>       
>>   Maybe Java OS might be more specific and practical :) . Anyway ,it is on my
>> ToDO list.
>>     
>
> You are late (at least on AVR): 
> http://www.harbaum.org/till/nanovm/index.shtml
>   
   nanovm ? :) that's too hardware dependent.  But the overview concept 
is approaching my idea . Thank you very much.
>   
>>>>    Now I saw your idea, I guess you mean JIT on AVR32 ?
>>>>     
>>>>         
>>> Well, in "Java Technical Reference" from Atmel they call it "Java Extension
>>> Module (JEM)".
>>>   
>>>       
>>   if you wanna do hardware acceleration ,we could consider libJIT which is
>> mentioned Kaffe document.
>>     
>
> Well, I'm not an expert in this, but I don't know if this really would 
> make any improvement. With hardware acceleration you "just" throw your 
> bytecode at the CPU and it processes it just as if it were native code. 
> So, I don't see how JIT could be useful here. Also, it is unlikely to make 
> our work easier, or would it? I'd go as simple as possible to start with.
>   
   You mignt mean Jazelle like technology actually. 
>   
>>>> I've been as a JVM developer for mobile & CE devices. Maybe I would be
>>>> glad to help you make you do a port to AVR32 ,but I am in China,the AVR
>>>> board might not be easy to find in public market ..
>>>>     
>>>>         
>>> That sounds very good! Now, after I've written my first email to the kaffee
>>> ML, I found JamVM, which is smaller, and is already ported to AVR32! In pure
>>> software mode, of course. So, I think, we could start with that one. If you
>>> prefer kaffee, you are certainly most welcome to try to port it over too! We
>>> anyway have ti first get it run in software mode, before implementing
>>> hardware acceleration.
>>>   
>>>       
>>    Yep,both Kaffe and JamVM are good thing to learn. What kind of applications
>> would you want to run on  AVR32?
>>     
>
> Ideally none:-) As I said, I'm just interested in the idea per se. Will 
> see how it comes out.
>   
   Hardware accelerated jvm for embedded device would be promising and 
much interesting . We could see on that day.
>   
>> On low-end devices,enable JIT would be expensive.
>>     
>
> Ok, then we better forget it (for now at least).
>
>   
>>    Now that JamVM is working on AVR32,maybe we need some benchmark or
>> something.
>>    I have said that I would be glad and interested to help a fresh port on
>> AVR32 ,so i think  JamVM would be fine..:)
>>     
>
> Good! Then I would suggest to move further discussion to the jamvm list 
> (https://lists.sourceforge.net/lists/listinfo/jamvm-general). I'll write 
> there with CC you and attach a very slightly cleaned up avr32 patch for 
> JamVM from Leen Toelen.
>
>   
>>> As for the hardware - as I said, as soon as I get the board, I can provide
>>> an ssh login to it to you or any other interester parties. I could even
>>> provide a serial console, if that will be needed too. I think, this should
>>> be enough for the beginning.
>>>   
>>>       
>>  That would be great....  There should be many experts here.....:)
>>     
>
> I'll let you know when the board arrives.
>   
   Thank you :) . Maybe I should find some emulator first :)
> Thanks
> Guennadi
> ---
> Guennadi Liakhovetski, Ph.D.
> Freelance Open-Source Software Developer
>
>   

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://kaffe.org/pipermail/kaffe/attachments/20080508/f8012e96/attachment-0003.htm