PPC976: Possible Xbox2 processor?

[Deadmeat]

...

why are we talking about cell in a thread about xbox ?

Because Sony fans are dragging CELL into it. You know the routine, "Behold! CELL the almighty is so great that MS dropped X86 like a bad habit and licensed CELL from IBM! CELL will prevail!"

XCPU2 is some kind of a Power5 derivative, and has nothing to do with CELL. Now, lets cut the CELL out and return to Power5 discussions.

 

[Paul]

Lose the cache and the core size drops to around 120 mm2 @ 90 nm.

And watch as efficiency drops like a tank, the cache is there for a reason not to just increase die size and money.

PSX2OAC = 8 watts @ 300 Mhz
PSX2OAC = probably ~30 watts @ 1 Ghz
PSX2OAC x 8 times the execution units = ~240 watts.

See, my 3 watts per APU was a very optimistic guess, because I don't believe the actual CELL APU will hit such number.

...

 

[Deadmeat]

..

And watch as efficiency drops like a tank, the cache is there for a reason not to just increase die size and money.

SMT takes care of that. Beside, XCPU2 is not running a database server like most Power5 systems will be doing, so it can with away with smaller cache.

 

[YeuEmMaiMai]

well it looks like Nintendo and Sony are going to maintain backwards compatability with their next consoles so if x-box is to do the same, they are going to need a CPU that is really really fast so they can emulate an x-box or it is going to be some form of a pentium X core........

 

[Paul]

SMT takes care of that. Beside, XCPU2 is not running a database server like most Power5 systems will be doing, so it can with away with smaller cache.

You didn't say smaller cache.. you said lose it.

 

[Deadmeat]

Lose the 2MB L2 cache and L3 controller and the overall chipsize shrinks to 40%. XCPU2 will do well with a L2 256KB cache instead.

 

[Psychogenics]

Lose the 2MB L2 cache and L3 controller and the overall chipsize shrinks to 40%. XCPU2 will do well with a L2 256KB cache instead.

Its almost garanteed that X2 will still be the biggest of the bunch.

 

[nobie]

well, they could put 2 more cores on the chip without touching the cache. @65nm that won't be such a huge chip.

 

[Deadmeat]

...

well, they could put 2 more cores on the chip without touching the cache. @65nm that won't be such a huge chip.

Power consumption wont' allow it even if space budget is available. Power5 is already running hot enough, having two more cores doesn't help.

Let me put it this way. Having a server grade chip inside a console is amazing enough, what more can you ask???

 

[Vince]

Re: ...

I feel intellectually violated having to read this, I'm sure I'm not the only one. Your responce to nobie was the last straw, this growing mentality of say whatever you feel like without proof or a sound known foundation and ignore all calls for validation is getting really old, really fast.

 

[jvd]

A quck question . Can they add cores in odd numbers or is it only even numbers .

 

[Deadmeat]

...

A quck question . Can they add cores in odd numbers or is it only even numbers .

I don't believe there is a specific restriction against odd increment in hardware design. However, most OSes do seem to require even processor increment, 2, 4, 8, 16, 32, etc...

So it is really an operating system issue and not a hardware design issue.

 

[Ty]

Re: ...

Talking to you is pointless, you have no knowledge. You have no training or education in the field (or else you'd said so when repeatedly asked about it), and you have no insight! You just make up to try and justify your already made-up, outlandish opinion.

Ya think?

Now I'm not one to call for a ban but this forum would be so much better if it had a rating system. One where its participants can vote on each other.

And heck, I don't even have a PS2 so I'm not much of a Sony fanboi!

The SNR is so low because of him imo.