Then please, do point us to those benchmarks.

Since I'm getting tired of this debate over common knowledge, I'll just quote Jon "Hannibal" Stokes of Arstechnica and Inside the Machine fame:

"At any rate, Playstation 3 fanboys shouldn't get all flush over the idea that the Xenon will struggle on non-graphics code. However bad off Xenon will be in that department, the PS3's Cell will probably be worse. The Cell has only one PPE to the Xenon's three, which means that developers will have to cram all their game control, AI, and physics code into at most two threads that are sharing a very narrow execution core with no instruction window. (Don't bother suggesting that the PS3 can use its SPEs for branch-intensive code, because the SPEs lack branch prediction entirely.) Furthermore, the PS3's L2 is only 512K, which is half the size of the Xenon's L2. So the PS3 doesn't get much help with branches in the cache department. In short, the PS3 may fare a bit worse than the Xenon on non-graphics code, but on the upside it will probably fare a bit better on graphics code because of the seven SPEs."

You obviously haven't grasped that SPEs are fast for some codes but slow for other. So AI is VERY taxing for SPUs unless you split the code to SPE and PPE dependend parts, which would add to the communication delays.

Secondly physics is not really the physics demanded in games. Game designers are not interested in how something works, instead they demand that the system works as they want it. So you use simplified codes, that can be easier tweaked and contains less unexpected surprises.

Furthermore: The cell has a dramatic problem with real physic applications (no direct double support and some applications have trouble with the limited amount of local memory in a SPE).

And if you really believe in benchmarks you only show that you don't really know what you are talking about. A benchmark only gives you a time for a fixed part of code with a fixed set of data. If you change the code or the data set you can get totally different results. So they have a limited value.