The article fails to mention the audience this is good news for: PC gamers. If next generation consoles are x86 based, expect to see future games being more widely available on PC, and, better yet, expect the "best" versions of those games (in terms of graphics, features, etc.) to be the PC versions. The only catch is if the game developers hold back on PC releases due to fears of piracy, but on the whole this probably will still mean many more releases on PC.
That said, most flagship titles (Halo, Metal Gear, Final Fantasy, etc.) will probably stick to a single console due to contractual obligations.
What? The instruction set matters very little. Pretty much all games are written in C++ and last time I checked there are C++ compilers targeting pretty much any architecture in existence. The graphics API is more important than the instruction set.
The Cell processor was very different, architecturally. I can easily see games originally designed for the PS3 to require significant effort to port to the 360 and PC. (While the 360 used a PowerPC processor that actually reused assets from the Cell, it was architecturally not that different from typical x86 multicores.)
For video games, the CPU architecture is still very important.
Remember when Halo came out on the Mac, and it was running slow as molasses on higher-performance hardware than what the game was originally written for? That's in part because many of the optimizations that make code run faster on x86 did the opposite on PowerPC. And any code that relied on SIMD instructions would likely have to be rewritten from scratch.
I'm not particularly familiar with the Cell architecture, but based on what I do know I suspect that going between it and x86 would be an even bigger headache.
The graphics API is more important for what? Gameplay? AI? Load times? Physics? Or did you mean rendering triangles, because indeed graphics APIs are very important for that.
I think it is extremely naive to assume that the CPU's underlying architecture and microarchitecture are utterly mundane details abstracted by the C++ compiler, ESPECIALLY in the case of consoles. Consoles are underpowered by today's standards, and making today's game run on them requires a deft maximization of both the GPU and CPU, not a mindset like "well C++ will take care of the CPU stuff". In case you hadn't figured it out, common C/C++ stuff from Havoc Physics to little old memcpy() and strcmp() are all hand-optimized for their platforms -- sometimes multiple implementations that make use of various ISA extensions. Try stepping through them in a debugger sometime.
Surely, /surely/ this is the least of their concerns. Since when did we get so sloppy as to assume byte ordering in files? When was the last time your PNG loader failed on x86 because the file has big-endian fields?
In the console programming world, cycles are precious. You don't tend to waste time doing things 'just in case', you're much more likely to assume a best-case scenario and engineer things that way.
In terms of endianness, it's not a huge problem - the toolchain normally copes with this, as assets are built individually for each target platform. This is what we did last time I worked on a cross-platform game anyway.
First Xbox was Pentium III PC with GeForce 3 Ti, all on single board in nice (that's debatable, but I liked it.) case, yet I don't think there were more ports than there are today.
I think it's gotten significantly better as time goes on. Going back to the SNES/Genesis days, ports (either direction) would often be an entirely different code base, so it'd be the same game only in spirit.
I also don't recall very many PSX/DC/N64 --> PC ports. MGS and FF7 would have been some of the few that I remember from those days.
It's definitely better today than it was 10 years ago.
Console makers are likely to continue subsidizing the hardware. So we'll just end up having cheaper playstations and xboxes.
I think very few people go out and buy a powerful graphics card. A console is simply a much more consumer friendly product that also happens to be a better bang for the buck in terms of hardware.
If the rumors outlined in the Ars article are true, there will be nothing to subsidize -- the console manuf are going for the Nintendo model after having their asses fed to them for 5 years while Nintendo turned a profit on 2-year old tech from launch.
It makes the most sense for Microsoft to go the compatibility-route a la Apple and iOS -- Apple proves it works, Middleware companies will love it (Crytek, Unreal, Unigen) and gamers won't need to skip a beat.
Here is an example, where the OS might lose to a console:
OS is like what the TSA is doing on the airports - it has to check everything, and it slows down quite a lot the queue of people going and flying. It's not that the airplanes are faster or slower (GPU, CPU)
Not the case on the consoles, where this is your responsibility, and your game is tested by QA, and later the console manufacturer itself until it passes. But no "TSA" there (or not to such extent).
For example (this might be a bit out of date, but used to be valid at least going back to Windows 2000) - You have a index buffer with 32-bit indices - on a console you know it's your responsibility to make sure the indices are valid, but no checks are done. On a PC, due to security restrictions the OS must make this check.
Final Fantasy XIII and XIII-2 have been released for both PS3 and Xbox 360. Maybe the game developers are starting to get the upper hand when it comes to contractual exclusives?
It's not that game developers are indentured servants (with the potential exception of first party studios like Naughty Dog or Lionhead, owned by Sony and Microsoft respectively), it's that the console manufacturers give big piles of money to good game developers in return for platform exclusivity.
So it's basically a business case thing, and a matter of leverage. Big, successful studios - say, Rockstar, makers of GTA - have a very strong bargaining position, and they know they can sell lots of units on all platforms. Whereas a weaker developer might be happy to take a boatload of money for an exclusive, as it de-risks the development somewhat for them.
The article clearly mentions that the xbox is still going to use PowerPC, and the playstation may mod the x86 processor.
And even if ps3 does leaves the processor as is, the OS they are going to use is not going to be windows. Perhaps a variant of Linux of some kind. This may mean more games on Linux (however highly unlikely), but definitely no new games for Windows PC.
I think you missed the point. The point is that if the processor is x86 based and the GPU is basically an off-the-shelf AMD GPU, porting the code from PS4 to PC will be a much easier process than it is today and make help game makers provide better feature parity between the console and PC version.
That said, most flagship titles (Halo, Metal Gear, Final Fantasy, etc.) will probably stick to a single console due to contractual obligations.