Originally Posted by Willynowei
You know what i'm going to exit the discussion, there's just too much misinformation here, its no ones fault, it just happens, i'm tired of trying to correct assumptions people have about things they don't fully understand.
There is a lot of comparison of specs by people who are erroneously comparing paper numbers without knowing the real significance of each figure. There is plenty of debate in developer and engineering circles, about the specifics of internal hardware and how the two systems compare. Whether one is more powerful than the other, or similar, or different, whatever the measurement, no one is sure of the exact gap right now in real world performance.
There's so many factors, and they are so complex. For example, there's so much to the Xbox's ES ram that you could easily write a whole book on the complex tradeoffs it entails for the system, from an APU workload-distribution efficiency boost to a latency boost in nanoseconds in comparison to a traditional memory layout. No one knows the voltage and heat of the memory and chip design, therefore, no one knows how high each chip can clock while providing adequate factory yields (even if the design recommends a 1ghz cap, you can have one system failing to reach the cap due to heat issues, etc).
There's no point in getting deeper and deeper into discussion when my general point is, we don't know. Lets wait and see. These companies have a lot of money invested to try and make successful entertainment devices, and powerful internals are a big part of that for both companies, even if its not their only concern.
The APUs are based on Jaguar and Radeon GCN silicon. There isn't a lot of unknowns for the hardware engineers about what this well proven silicon can handle.
Are MS and Sony both designing towards an energy and heat budget? Sure, but enough leaks have come out that we have a pretty clear answer as to what each is capable of.
Also, the latency "benefit" from ESRAM is measured only in nanoseconds and also largely irrelevant for anything other than changing code midstream (like you do on a PC OS when you close a program and open another). It has no appreciable benefit to single application coding.
It isn't erroneous at all to compare paper number of two x86 based APUs built by the same company using the same family of silicon as the origin. This isn't apples to oranges. This is apples to apples where they both fell from the same tree but different branches.