Speaking of which.. I should get some people (all like, three of y'all) to join my Playsyde folding team. It's a team of one, currently, mostly because I never bothered to tell anyone about it ;)

But sorry that's way OT, bash away!

I already joined a team. Left my PS3 on for 50hrs. And as sad as it sounds, something in me was too lazy to leave the PS3 on. . So I've stopped. Wow. Feels good to get that off my chest.

Seeing as graphics are the area of focus for this title (i.e. no CPU intensive/defining gameplay) I don't think we would have seen a title this this significant of a difference namely because the 360 GPU is more flexible. You are not going to find a lot of areas where RSX is significantly faster, let alone common ones that are significant deal breakers visually.

PS3 SC4 is a victim of a number of things. Obviously tools is one as well as lead SKU being the 360. But SC4 is banging on a couple of doors that are not friendly to the PS3 GPU. e.g. RSX can only do HDR through the shaders when using MSAA, Xenos can go this route or through native formats compatible with FP blending. It appears (as I am told) SC4 went the FP blending route so the boatload of extra work to redo the renderer wasn't really an option; but in the flip case with a shader based approach Xenos would have done just fine. Geometry is another issue. Xenos has a significant edge in geometry performance and is wider. This is why Sony's Edge software has a lot of culling and vertex work being done on SPEs. While you can work within RSX geometry limits just fine you are not going to run into an area where porting your geometry code to Xenos is gonna be unreasonable with a little work whereas vice versa may be significant redesign and workflow. Memory is another issue in SC4. On the PS3 you have about 386MB in a split memory configuration (512 - 96 OS - 30 framebuffer) whereas on the 360 you have about 473MB of shared memory (512-32-7). 20% less available memory is a one way issue. Add in the additional memory management of having to work in framebuffer traffic for performance as well as managing 2 memory pools and texture penalties and you just created a lot of work. And it doesn't stop there as there are compiler and tool disparities and little things--like scaling just works on the 360 and the PS3 takes more planning. And we didn't even discuss the disparity in online networks and the work involved there.

So while you can hold to your belief that if the tables were reversed the results would be the same, but flip flopped, I don't agree in graphic limited scenarios. It is absolutely true that there will be areas and games that RSX will do better than Xenos when designed for RSX-SPE workflow and then ported over. But there are a lot less "gotchas" -- RSX HDR formats will work fine, Xenos has more memory available with less management and possible penalties, Xenos is a geometry monster, Xenos won't be hit with fillrate performance issues, Xenos performs better at most PS SM3.0 code, etc. So where there will be deficiencies the developer can remain more focused on resolving the bottleneck or just simply disabling the graphical effect.

SC4 is for sure a more extreme example, but if you go and compare most of the multiplatform titles to this point most of them are graphically better on the 360.

And it tells you when like Insomniac pretty much admits Xenos is "slightly" faster.

Where the 360 will run into problems is quite different. Games that the 360 is going to choke on would be built around things like

- Use a lot of optical storage (BDR > DVD9)
- Require a HDD for storage or fast caching (HDD standard > non-standard)
- Use motion sensing (Sixaxis > 360 pad)
- CPU bandwidth limited scenarios (25GB/s peak with another 22GB/s for RSX versus 22GB/s total for the CPU to share with Xenos texturing)
- Gameplay mechanics that are dependant on algorhythms that map effeciently to Cell (Cell math advantage as well as architectural perks like very fast bus and insanely fast local memories for each CPU)

Of course these advantages are harder to tap and even harder to develop gameplay around. They are either expensive (filling BDR), difficult and labor intensive (getting ones head around Cell and fully exploiting it), or issues developers have struggled with conceptually (HDD killer app, motion detection).

This is contrasted with, "Hey, over here guys we just give you 20% more effective memory, better tools for performance tuning, eliminate a number of bottlenecks so you don't even need to worry about things like fillrate, and are design is flexible to your workloads be it pixel or geometry work". Those are not sexy things to consumers but they do make programming a lot easier.

Whatever one wants to peg these issues on (it is a sum of many) the reality is that Sony is in a tough situation going forward. A lot of games are looking better on the 360; further the PS3 is too expensive and losing mindshare due to losing ground in exclusives, lack of early killer apps, Wii's run away success, and MS building a significant next gen portfolio. Sony needs to convince publishers to invest in the PS3 so its advantages -- CPU power, BDR, motion sensing, standard HDD, etc -- can become selling points. But to get publishers to make that investment you need an install base. And an install base is contingent on price and killer apps. And to get the killer apps you need to convince publishers to exploit your system which... uh oh.

Sony was really leaning on BDR to be its killer app. Further ATI did an excellent job on the 360 graphics systems and MS bit the bullet and went with 512MB of memory instead of 256MB. If the 360 had an X1800 GPU and 256MB of memory we would be looking at a different game all together. The PS3 would be superior technically across the board and with little optimizations would be able to do what the Xbox did to the PS2.

How each company plays their cards in 2007 will be important. One problem Sony has is they are really lacking in the games and MS is loaded -- and that is an understatement.

So even if I am absolutely wrong about what issues SC4 had (heh heh) that isn't the real issue. The real issue is:

What is Sony going to do to ensure future 3rd party multiplatform games don't end up like this again, and importantly, begin differentiating themselves from the 360 and showing off how the PS3 is MUCH better, $200 better. The early adopter season is over, now it comes down to compelling reasons to buy. And publishers are watching this. Sony bet a lot on "we can sell 5M with no games" philosophy in sticking with the PS2 and banking on the fact they had a near monopoly on developer support and had Japan and EU won before the game started. Sony didn't quite get those sales and the benefits of their platform are long term whereas their immediate issues are short term ones more difficult to resolve (ease of development, price). I would really hate to be Sony execs right now. PS4 on the other hand is sitting absolutely pretty. I think MS may be screwed in regards to their X3 CPU and there won't be any more Xenos tricks to pull out of the hat next time as RSX-2 will go eDRAM and with GPUs more performance oriented we won't be seeing many more huge paradigm shifts like unified shading. And it looks like BDR is gonna win so MS is gonna have a BDR in the X3. What goes around comes around...