AMD/ATI for Xbox Next?

Just because developers want something doesn't mean it is the best implimentation of hardware. Mind you developers thought faster serial processors was the best path as well as beefier single cores.

Just because, in theory, a single processor would produce best/easiest results, doesn't mean that is the best use of the hardware.

e.g. Repi mentions a OOOe being a dot in a sea. Maybe this same concept can be achieved through a very fast link between a GPU and CPU.

Anyhow, my problem is first that while it sounds great in theory, we aren't seeing any benefits of a single CPU/GPU now and we DON'T know how it stands up against a current CPU GPU. And the 2nd issue is that a single chip has fewer execution units for the same $.

So you pay the same $ for lower peak performance and no one has shown that the advantage of a single die can overcome this issue to be at least on par. That is a big challenge--and there is always the counter arguement that why cannot a 2 die solution evolve (e.g. optical link) to get the same concept without the sacrifices of a single die.

 

having a single gpu die in the cpu could be used for power saving while playing back 1080p video content and other things. That way you'd only have the cpu active in such a set up.


They should still go with something dedicated though. I'm thinking a hybrid of the next ati gpu would be good. it be a second generation dx 11 gpu with perhaps some foward thinking tech for dx 12. But being a second gen dx 11 gpu it should be better at doing all dx 11 things.

Then perhaps a 6 or 8 core bulldozer chip ? With the tesselator they should be able to do alot more stuff in the gpu and reduce the load on bandwidth and the cpu for these things.

 

http://arstechnica.com/hardware/new...g-larrabee-route-on-road-to-cpugpu-fusion.ars

Some more Fusion info/overview.

 

Does I understand right and the guy thinks that AMD response to larrabee and aim with fusion is bulldozer core?
I think he is mislead... he could also say that Sandy bridge will kill larrabee purpose with 8 cores and 256bits SIMD
units...

 

I believe what he was trying to say is that AMD's "future end-goal" for Fusion is something Bulldozer-like. Note how he indicates such a future is kind of a flip flop in execuation units it appears.

 

I could see a Fusion mark 2 end up in the next Xbox.
Backwards compatibility might not be too bad, due to the 360's reliance on higher level libraries and intermediate code, as well as the low performance per thread. If a game were to make use of all 6 threads, you might get performance per thread on par with the original Xbox cpu, which could probably be emulated by current top end cpus. (the wii's cpu is of similar performance, and is already emulated at full speed by top end i7s)

I'd like to see Microsoft pull back from the hardware of Xbox and focus on it more as a software platform.
Unify PC and Xbox game development, so every Xbox game is capable of running on a PC of sufficient specs. It would also allow MS to charge a license fee for every PC game sold, but considerably boost PC development. Microsoft could even purchase off the shelf components for the Xbox, and just upgrade to whatever is comparable in performance but cheaper over time.

They could even license out Xbox as a brand to be produced. MS could produce the base gaming Xbox, and could even have somewhat conservative specs. Then Dell and others could produce higher end Xboxen, possibly with additional functionality or just higher performance. Yeah yeah, it's the 3do model, but it could work. Kill off Windows Media Center and replace it with Xbox media center. Media centers would no longer be crippled by running a traditional PC interface, allowing the hardware and software to be better customized around a controller. Think AppleTV, it still runs OSX, but not in any way exposed to the user (unless hacked).
And PCs would still be PCs, but able to play Xbox games. PC only games could still exist, if devs had no interest in targeting the Xbox platform.

 

Seconded... Nothing else need be said.

 

And a "Fusion" type CPU but with PowerPC cores?
IBM and AMD have many accords, and I have got the feeling that their strong partners?

 

Thirded!

There's a reason why consoles were invented in the first place. If MS just turned console gaming into PC gaming then we'd be moving backwards and not forwards imho.

 

I guess the question boils down to how easy/hard it would be to combine the PPC core with an AMD GPU on the same die and whether it would be better than an x86 solution, better in this case meaning less $$$ per chip for MS considering two equally performing CPUs.

 

Yes, you right and also better in the sense of perf/watt and perf/$$.

 

It looks like early speculations about die size and Sp count were wrong anyway this is kind of interesting:
AMD X86 is now royalty free
It may remove a "legal burden" if AMD were to let MS buy some fusion IP, what do you think?

 

AMD doesn't have to pay any royalties to Intel and vice versa, but thats a long way from allowing AMD to extend X86 licensing rights to other companies.

 

I'm not saying anything about console gaming needs to change, other than software compatibility with PCs. Why not? Xbox and PC already share many libraries, and .net is already a high performance just-in-time compiled language, Microsoft could increase the appeal of the Xbox platform to developers if it also produced PC compatible code.

That's practically what PC gaming is already, except developers have to do a minimal porting effort in order to create the second PC sku.

 

OK, I went a bit too far.
I read a thread about new AMD cores (bobcat/bulldozer) and some of the persons thank that bobcat could be what they call a "softcore" ie it could be customizable by the costumer. That's close to what ARM does/did. I don't understand properly the in & out of AMD/Intel agreement tho...
But people having this discussion have some clues to say the least, so I feel like bobcat could an option for MS.

 

Eeeeexcept you added other steps - 1.) Modularizing the internal componentry of the console, making it fully upgradable and breaking any target developers have to aim for, (2.) while at the same time introducing a dozen or more user-side horrors. 3.) Licensing the hardware out to be modified further by vendors - now instead of three consoles, we have three dozen... all with the same name, and completely different spec and performance. "My new xbox game is broken!" No, it just doesn't work with that xbox... "Why? It's a new Xbox!" Because it's a new, budget-bin DELL Xbox with 2GB RAM, Core2Duo, and a Radeon 4330... It's a bad idea, and as a hardware company, 3DO died for it.

We're already producing software that at compile time differentiates code into several hardware-specific binaries, though - that part is already here, but I think that's as far as it needs to go. The rest of the confusion can stay PC-side.