Mr Bean wrote:
To add another issue, why do you think Corridor shooters are corridors? Or why destructibility is so limited? Ram limitations factor into both of those.
Actually the cost of content creation is the main limitation for the former and the lack of sufficient physics processing horsepower is the main reason for the later. Consoles can stream large worlds as demonstrated by Just Cause 2, GTA 4 etc, memory limitations mainly cause repetition (only a few kinds of car on the streets at once) and some pop-in on geometry and textures. For debris you run out of CPU/SPU power way before you run out of memory (particularly on the X360, which has twice the usable memory of the PS3); however memory limitations prevent destruction from being persistent
. I'd note that 'corridor shooter' as a design pattern actually goes beyond even content limitations; if you look at say Duke Nukem 3D's levels, they are more of an interconnected network of rooms and areas instead of say Mass Effect 2's cunningly disguised corridors (or Skyrim's completely undisguised ones), even though the later are much larger. A lot of the corridor shooter trend is game designers just not trusting players to have spatial skills or not complain about backtracking or bother looking at a map. Dead Space is a little better about this mainly because the 'show pathing on my HUD' feature made the designers a little less anxious about players getting lost.
Super long load times are another factor plaguing console games. A PS3 hack was done awhile back And reported by Toms Hardware
that by adding an SSD speed you cut average loading times by half. And that was a hack job by some enthusiasts. If you had it done professionally and with a decent amount of ram and smart pre-loading to back it up you might see the PS4 or next Xbox loading games three times faster than the current gen.
The important thing is not the bandwidth as such, it is the ratio of bandwidth to system memory, as developers generally size content to fill most of system memory (with a bit left over for streaming where applicable). The problem with SSDs is that they are still too expensive to be compatible with the idea of gamers downloading lots of full-size titles and HD movies, which will easily fill up a terabyte HD. A hybrid solution is possible, but somewhat shorter loading times are not as good a feature for the console launch hype machine as better graphics or gameplay. Fortuantely this is something you can differentiate between models and upgrade later in the console lifetime without breaking compatability.
The PS3 and Xbox 360 were very much hack jobs, done just as hardware was refreshing a cycle and there was a hundred and one improvement that could have been made. The PS3 was released in 2005 with an early 2004 GPU with only 256 megs of ram. The Xbox was just as bad, fancy CPU's or no the rest of the stuff in the box was lackluster for 2005 even then.
The PS3 was a hack job caused by the fact that Sony originally wanted two Cells. When the Cell turned out to be completely incapable of rendering 3D graphics at anything vaguely like competitive performance (and pretty useless for most other tasks, but that's another story) they were forced to grab a GPU off the shelf and shove it in. The Xbox360 on the other hand was an excellent hardware design for the time. The unified shaders were very forward-looking and gave developers a lot more flexibility in shader loading, the unified memory was vastly better than the PS3's split memory, the VLIW5 design was fundamentally much more efficient than the contemporary Nvidia shader cores, the CPU design was more like a hex-core PC than the 'PS3's dual-core PC with broken ridiculously hard to use DSPs glued on', the dedicated video scaler chip further unloaded the GPU and improved output quality over the PS3. Most of all the eDRAM was extremely useful and saved a lot of the most precious commodity, GPU-memory bandwidth; depth/stencil test and alpha blend are almost free on the X360, MSAA is much cheaper than PC or PS3, plus it saves a little system memory (storing the working frame buffer). Developing for the X360 is a vastly more pleasurable experience than for the PS3 with the sole exception of when you absolutely need more physics engine performance, at which point the raw FLOPS advantage of the SPU array is worth the pain of using it.
The only serious issue with the X360 hardware (in the late 2000s) was that it didn't include a high-capacity optical drive, and that's only because Microsoft weren't motivated to burn piles of money to win an optical format war. If it had a lifetime of four years like the original we'd have no cause for complaint, the huge drag on game development progress only developed because the manufacturers were determined to stretch out the console lifetime to twice as long in this generation.