This is one of the most uneducated looks at the next gen consoles. There are so many mistakes I can't even BEGIN to list the errors, let alone the lack of knowledge of game engineering. A few glaring errors:
The SPEs of the Cell processor are in fact fully fledged 128-bit SIMD RISC processors. They are based on the PPC instruction core, optimised to be fed by a constant flow of data. They have no cache as such, but have a 256KB local store, which operates at the full processor bandwidth, negating the need for a cache. They also have access to main memory, via the EIB, at 3.2Ghz, again negating the need for a cache. The SPEs can perform 4 x integer, 4 floating point, or 2 x double precision floating point each clock cycle, making them as fast at integer operations as floating point. The SPEs can be fed with a constant flow of data from main memory without needing a cache as the memory operates at the full bandwidth of the processor.
This in fact makes it far more suited to games than a general purpose processor: An general purpose CPU is suited to a desktop processing environment because it has to swap between many different tasks - of very different natures. However, a games console generally performs the same tasks continuously - the cell allows each task to run on its own core and do its job without interruption. The Cell in the case of the PS3 can run 9 simultaneous threads - 2 cores on the PPE and the 7 enabled SPEs. All can run the same machine code. Any dynamic changes in procesing will be handled by the PPE, leaving the 7 SPEs to carry on number crunching
While this was blatantly a pro MS article, the author still managed to get wrong the 360's details - the Xenon processor in the 360 is actually 3 dual core units, giving 6 processor cores all together, however these are optimised for a desktop environment, not the constant numbercrunching of a games console.
Both are made by IBM, and given that IBM chose to use 16000 Cells in their newest supercomputer which will be capable of 1 petaflop (1000 Teraflops) and not the Xenon, says just which is the higher performance processor.
I'm not going to go through and correct the further errors of the article, I feel I have written enough. I would encourage the author to do further research rather than writing biased wishful thinking.
