Hi,

There is a simulator available from IBM for the Cell processor.
Unfortunately it runs on a Fedora Core 4 x86 system, but it is pretty cool.
It is a full system simulator and runs either standalone programs or a
version of FC4.  There is also a PowerPC 970 simulator available.  I have
played with the Cell simulator and it is pretty cool.  I have not tried the
970 simulator yet.  More information and links for the Cell simulator
downloads can be found here:

http://www-128.ibm.com/developerworks/power/cell/

And for Power Architecture in general and the 970 simulator:

http://www-128.ibm.com/developerworks/power

The 970 simulator could be useful in the porting effort.

Lloyd Staley


On 11/27/05 4:32 AM, "Sven Luther" <sven.luther at wanadoo.fr> wrote:

> On Sun, Nov 27, 2005 at 04:43:20AM -0700, William Kucharski wrote:
>> 
>> On Nov 27, 2005, at 2:57 AM, Gabriele Bulfon wrote:
>> 
>>> There's a lot of rumor about the Cell Processor being researched by
>>> IBM/Toshiba/Sony for their next PS3 platform:
>>> http://arstechnica.com/articles/paedia/cpu/cell-1.ars
>>> http://www.blachford.info/computer/Cell/Cell0_v2.html
>>> 
>>> What about Solaris 10 on such a platform?
>>> What the goods and bads?
>> 
>> The biggest drawback is that Cell is NOT designed as a general purpose
>> computing processor, but as a very specialized processor for video
>> game usage.
> 
> Notice that :
> 
>   1) the Cell processor has 8 (or 16, not sure i remember well) Cell units,
>   which are capable of handling their own (DSP or SIMD like) instruction set,
>   which is totally unrelated to powerpc.
> 
>   2) In addition of the Cell units, it also has a reduced PPC970 core to act
>   as control processor, and for handling non-purely-graphic stuff, like game
>   AI or other such. This is also the one booting the initial system.
> 
>   3) The PS3 is interesting, but more interesting will be future IBM based
>   Cell blades, using the Cell units in scientific processing and such.
> 
> As such, it should be relatively easy to run Polaris on the Cell, but taking
> advantage of the actual Cell units would maybe not be that evident, and it
> should be slower than an equivalent PPC970.
> 
> Friendly,
> 
> Sven Luther
> 
> _______________________________________________
> powerpc-discuss mailing list
> powerpc-discuss at opensolaris.org


Reply via email to