Squeak on the ... Xbox 360?

Hey everyone!

Another thing I'm wondering- has anyone tried getting Squeak going on  
the Xbox 360? Like the PS3, it's an incredibly powerful machine-  
three async multi-processing cores at 2.5 GHz, if I'm remembering  
correctly.

For anyone interested, you can download the XNA Game Studio Express  
1.0 for free from MS's site at the URL below. Hobbyists can use to  
develop programs for the 360 without having to use a modchip or other  
DMCA-violating measures.

http://msdn2.microsoft.com/en-us/xna/aa937795.aspx

I'm thinking about getting a 360, and probably will- I'm just waiting  
around hoping for a drop in price.  If no one else has tried it or  
plans on it, I will probably take a whack at it. Now that I have an  
Intel Mac Book Pro, I can run these Windows-only programs!

Regards,
Aaron

2007/1/28, Aaron Reichow <[hidden email]>:
> Hey everyone!
>
> Another thing I'm wondering- has anyone tried getting Squeak going on
> the Xbox 360? Like the PS3, it's an incredibly powerful machine-
> three async multi-processing cores at 2.5 GHz, if I'm remembering
> correctly.

PS3 is about half the speed of a 1 GHz PIII for Squeak. I think it is
reasonable to expect the Xbox 360 being in the same range for Squeak.

Philippe

Philippe Marschall wrote:
> PS3 is about half the speed of a 1 GHz PIII for Squeak. I think it is
> reasonable to expect the Xbox 360 being in the same range for Squeak.

Where does this number come from? Given that PS3 utilizes a POWER-Core
at 3.2 GHz it sounds somewhat unreasonable.

Cheers,
   - Andreas

2007/1/28, Andreas Raab <[hidden email]>:
> Philippe Marschall wrote:
> > PS3 is about half the speed of a 1 GHz PIII for Squeak. I think it is
> > reasonable to expect the Xbox 360 being in the same range for Squeak.
>
> Where does this number come from? Given that PS3 utilizes a POWER-Core
> at 3.2 GHz it sounds somewhat unreasonable.

http://lists.squeakfoundation.org/pipermail/squeak-dev/2007-January/112829.html

Philippe

On 28-Jan-07, at 4:00 AM, Philippe Marschall wrote:

Clearly there is something odd going on with the combination of  
compiler/vm/cpu on Jerry's PS3 since a 2GHz G5 mac gets 157441574  
bytecodes/sec; 6467114 sends/sec. Since the bytecodes figure scales  
amazingly closely to clock speed I'd expect more like 230mBc/s and  
perhaps 8m sends/s since that depends a bit more on cache and memory  
arcitecture. Assuming of course that the claims I've read of the Cell  
having a 3.2GHz POWER core are correct.

The xb360 also has 3.2GHz ppcs so I'd anticipate similar performance.  
With 3 cores it might even be possible to run 3 copies and get decent  
performance if one is lucky enough to catch the process scheduler in  
a good mood and able to run the main threads at one per cpu. No idea  
if the xb can even do that.

tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
Adding manpower to a late software project makes it later.

On Jan 28, 2007, at 18:43 , tim Rowledge wrote:

> Assuming of course that the claims I've read of the Cell having a  
> 3.2GHz POWER core are correct.

Well, *I* read that it's not even remotely comparable to either a  
POWER or a PowerPC core, it just happens to use the same instruction  
set. It's stripped to the bare minimum, for example, it does not do  
out-of-order execution. Or, while a G5 can issue 8 instructions at a  
time, the Cell does only 2. I can't find a good reference right now,  
but a bit is mentioned here

        http://arstechnica.com/articles/paedia/cpu/cell-2.ars

and here

        http://www.blachford.info/computer/Cell/Cell1_v2.html

- Bert -

That's my working theory too.   John McIntosh has offered to help me  
check out various VM issues but I've dropped the ball on my end.  I  
hope to free up some time to follow up this week.

Jerry Bell






On Jan 28, 2007, at 12:53 PM, Bert Freudenberg wrote:

Bert Freudenberg wrote:
> On Jan 28, 2007, at 18:43 , tim Rowledge wrote:
>
>> Assuming of course that the claims I've read of the Cell having a
>> 3.2GHz POWER core are correct.
>
> Well, *I* read that it's not even remotely comparable to either a POWER
> or a PowerPC core, it just happens to use the same instruction set. It's
> stripped to the bare minimum, for example, it does not do out-of-order
> execution.

IF true, this would suck. Do we know if the xbox has the same limitations?

Cheers,
   - Andreas

2007/1/28, Andreas Raab <[hidden email]>:
Wikipedia says yes:
http://en.wikipedia.org/wiki/Xenon_%28processor%29

Philippe

> Cheers,
>    - Andreas
>
>

Don't forget Apple did evaluate the Cell for "general purpose  
computations" when they were evaluatiing going forward with PowerPC,  
versus Intel or AMD.

And decided it wasn't general purpose enough, ie performance wasn't  
what you would expect, mmm wonder where the reference to that study  
is...

  I'll note the core 2 duos crank out some impressive numbers, likely  
all parts of the VM needed to do the tiny benchmarking get sucked  
into  the primary cache.


On Jan 28, 2007, at 12:54 PM, Andreas Raab wrote:

--
========================================================================
===
John M. McIntosh <[hidden email]>
Corporate Smalltalk Consulting Ltd.  http://www.smalltalkconsulting.com
========================================================================
===