From Gamespot.News: “… Kutaragi (Sony Computer Entertainment president) disclosed that he plans to install the Linux operating system on the PS3’s hard disc drive (HDD) so it will be recognized as a computer, rather than a mere console.”
Playstation 3 will also run Linux
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
BeOS is multithreaded all the way through, the opensource version, Haiku, could be made to run on PS3, imagine the SPE’s being fully used 100% of the time
Nah, 100%, who is using their system to the max al the time anyway? Yeah gamers : )
http://www.haiku-os.org
BeOS is multithreaded all the way through, the opensource version, Haiku, could be made to run on PS3, imagine the SPE’s being fully used 100% of the time
That’s all well and great, but still doesn’t help with dealing with SPE’s local memory.
But never mind the Cell, does Haiku even support bog-standard symmetric multi processor (SMP) machines? The website doesn’t say.
Now I get to recompile the kernel every time a new game comes out!!!
i could have sworn that every SPE on the cell have its own (alltho small) cache.
ah, found some informative reading:
http://arstechnica.com/articles/paedia/cpu/cell-1.ars/2
if i understand the text right, every spu/spe have about 256k (hmm, no b or B so i cant tell if its bits or bytes
) of dedicated storage.
and check out the picture here:
http://arstechnica.com/articles/paedia/cpu/cell-2.ars
the power core at the top have a L1 cache, there is a general L2 cache, but see that every spe is hooked into that eib? from what i understand this allows every spe to request data directly from the main ram without having to go via the L2 or anything else.
I was reading arstechnica too, it’s a great site.
every spu/spe have about 256k (hmm, no b or B so i cant tell if its bits or bytes
) of dedicated storage.
That’s right, but it’s local storage, not cache. The software has to fill and use it explicitly, i.e. existing software wouldn’t benefit.
from what i understand this allows every spe to request data directly from the main ram without having to go via the L2 or anything else.
Yes, the SPEs can access either the L2 or access main memory directly.
Bypassing the L2 makes sense for streaming data that isn’t going to be reused soon anyway. Otherwise the SPEs would still want to go through the L2 because main memory usually is a lot slower.
oops, i just spotted a small () comment in you post about compiling/gentoo where you specificaly talk about getting gcc to use the spe local memory.
looks like my post(s) where a bit useless…
still, i wonder how much work it would be to get gcc to use the local mem on those SPE’s.
hmm, isnt there a way to spread a compile over several seperate computers? could a similar trick be used but have each SPE act as if it was a seperate computer?
hmm, isnt there a way to spread a compile over several seperate computers? could a similar trick be used but have each SPE act as if it was a seperate computer?
Good idea, but I suspect gcc needs a lot more than 256kBytes.
heh, just stream it of the main ram
point is to spread the workload into chunks and then put it back together in the end