You are utterly revelling in your ignorance. First of all, Multiple cores are absolutely useless in emulating a single CPU, period You can only use ONE of those 3 cores. Second of all, you are using the absolute maximum theoritical performance, something that simple will not happen. Realworld performance wise it won't come close. In fact you can be quite certain that the P3 will come closer to it's theoritical max than the X2-CPU will.ggs wrote:3ghz * 2 instructions per cycle = 6*10^9 instructions per second. With 3 cores, thats some 18*10^9 instructions per second, theoretical max.Praxis wrote:Even if the 3 GHz PPC only gets 2 calculations per clock cycle?
You're being terminally retarded to talk of your ass. You have no idea what you are talking about. The much shorter pipelined out-of-order and wider P3 will likely have a similar if not superior IPC, and in some case will simple OWN the PPC because of it's out-of-order nature. Learn what out-of-order execution means, it's history, etc. It's a very important aspect of a CPU.Compared to the p3 733mhz processor which is considerable slower.
You would have to be terminally stupid not to be able to emulate a processor when you have 3 times the number of processors and probably close to 3 times the raw preformance.
True, but you still need the CPU.Most of heavy lifing for Xbox games gets offloaded to the GPU anyway.
.....
Now you really pulled this one out of your ass. the P3 is NOT a netburst architecture. Next time get a clue. Irregardlessly, a modern Netburst based CPU will, in fact, asskicked this PPC in IPC because the IPC of the PPC is just that bad.Nope. Tried doing some googling for it but no luck. But considering its a intel chip based on the netburst tech, then it cant be that great.
This is ridiculously over inflated. I'll love to see how they'll handle x86's custom 80-bit floats with a VMX unit that only supports 32-bit floats with anything near 70% efficiency. Emulating at these performance levels is ridiculous. The closest you can get is perhaps 50% loss of performance, and thats with a method of translating x86 instructions and then caching commonly used translated x86 instructions based an x86 emulator for Alpha chips, but this may not apply for PPC. A straight emulation will be much slower. Given the fact that the PPC is not that much faster than a P3, this is not looking up.JIT compilation of the x86 code into PPC is also doable too. As for JIT preformance, MS .NET has ~80% to 90% preformance of equivelent native code for the most part.
Even if the JITed x86->PPC code was 70% of the original, then the extra speed of the PPC core easily makes up for it.