My first thought was, "That is a 360 shot... has to be because there is no AF on the left part of that wall!"

HAHAHA

Texture bandwidth was a theory ("championed" will you by a very strong PS fan) but Mintmaster and Dave seemed to put the kabosh on that ideas as IMO, especially in regards to AF bandwidth needs (which are nominal).

Further, lets compare this to a 7600GT, which also has 22.4GB/s of bandwidth. The 7600GT doesn't have significant problems running a game like Quake 4 at 1280x1024 with 4xAA and 8xAF -- nearly 70fps. The 360 is only rendering most games at 720p @ 30fps. While the 512MB is shared with the CPU, high end PCs have ~6.4GB/s of bandwidth and most games don't push that even. Anyhow, so while the 360 takes a hit (lets say 6.4GB/s) for the CPU, leaving 16GB/s of bandwidth for texturing, the 360's memory is not impacted by color, z, alpha, etc as these are ALL done on the eDRAM. So much of the bandwidth hit taken from CPU usage is easily reclaimed through all the backbuffer, the biggest bandwidth client, all being isolated on the eDRAM.

PC GPUs have been doing AF with very minor penalties for years. ATI has been doing this since the 9700 back in Fall 2002. So

* AF does not require signigicant system memory bandwidth
* The 7600GT has less effective bandwidth than Xenos, yet 8x AF is routinely used with marginal penalty
* AF in general has been a minor penalty for years

That leaves a number of other alternatives

* Xenos is broken. Unlikely for a number of reasons, one primary one is AF IS used at times
* K is not mature yet in this regards
* Developers are struggling with stable framerates (see how GRAW turns off Vsync at times) and every extra frame or two is worth it; combine the rushed deadlines...
* Dev time... last feature turned on, but cannot be done globally enabled since it needs to be turned on by surface
* PC devs are lazy and just used to throwing the switch

Who knows. It could be a number of reasons, but right now no one with inside info is talking, and a lot of the conjecture is pretty weak.

What seems to be pretty well established though is games are not using it universally, but it DOES appear to be used to a degree. It is annoying (I love AF) but jumping to the conclusions some do is hasty. Of course with E3 and the PS3 around the corner expect more and more of this.

Yeah, that is a problem devs are still cautiously working with. Destructable worlds mean some big problems

* Players can destroy key NPCs, assets, etc... and block/destroy objectives, thus destroying scripting and gameplay
* Accurate deformation takes a lot of horse power and geometry manipulation whereas it seems devs have gone more toward the pixel shader "fake it" route with a lot of faked geometry
* Physics is still in its infancy where they are mainly "physics effects" and few "physics gameplay"
* Lighting/Shadowing models need to be pretty much completely dynamic on all objects to look good and dynamically adjust to the destruction and havok the player creates... of course unexpected destruction could cause all kinds of glitches and bugs
* Play / Bug testing such games would be very, very difficult with a very questionable end product in terms of gameplay
* Totally free and dynamic deformity would not work for online games (too much information to pass over the internet), especially for physics-gameplay effects where objects that are destroyed remain as such. Ruble shape, position, weight, etc... x 10,000 objects would bring servers and broadband connections to their knees

At least those are the arguements I have heard. Personally I hope someone like the Red Faction devs give it a go anyhow. I want completely open ended gameplay!

So it made the building vibrate..but did not damage it :P