http://www.gametrailers.com/player.php?id=9263&typ...

multiplayer video, showing off the wall walking "path's" and "spirit walking" (ie your spirit leaves your body)

looks to be an awesome game on- and offline....

Something we can agree on ;) Smart marketing, but still misleading. NV had actually released some slides of how "1080p and AA are here now" but when you take a closer look what they did was list a TON of old games to obscure the impact on games like Far Cry, Doom 3, Chronicles of Riddick, and Half Life 2 (i.e. the most "intensive" games at the time of the slide). A while back someone had taken benchmarks of the 7800GTX 512MB (basically comparable to the RSX from the E3 specs Sony released for the PS3 GPU) and it became very clear that (1) AA nor HDR were realistic at 1080p and (2) at 1080p modern games struggled to get near 60fps.

It is completely unrealistic to expect the E3 render targets (Motorstorm, KZ, etc) at 1080p, let alone with HDR or AA.

Inflatable also mentioned Epic and I would agree. 1080p is 2x the pixels of 720p. 1080i is not great for action games because it is really only 30fps (60 fields, but 30 frames). That means a 720p game can either use up the same graphical resources with 2x the shader detail at 30fps OR at 720p you can effectively double your framerate in GPU limited situations.

I work a bit with photographs and such, and one of the problems with super high resolutions at this time is that they do not increase the detail to the assets themselves. We are not so far beyond the age of low quality textures, repeated textures, hard poly edges, etc. 720p is a nice jump up for clarity, while still being a good compromise in detail. A good example of this would be if you could play the current gen FNR3 on a PC at 1600x1200 or play the Xbox 360 version at 720p. The Xbox version would be at a lower resolution, but the lighting, detail, animation, textures, anti-aliasing, and so forth look so much better on the 720p image that while the 16x12 is a higher res it does not look as good.

Considering how even the best GPUs are currently struggleing to break 60fps with AA and HDR (ir capable) at 16x12, I think 720p was a GREAT decision by MS seeing as it is a 5 year console and the fact most HD TVs are 720p or 1080i (heck, most 1080i are CRTs that contain at most 1400 lines anyhow, and many far less than that!) and 1080p wont be a consumer standard for many years. I would be surprised if it even had a significant market penetration (>15%) by 2011 at the next round of consoles.

I am not sure when the next consoles will come. Sony stretched this one to a full 6 for them. Nintendo is keeping 5. MS went 4 due to the fact the Xbox1 was not scaling is cost and knew they could jump on Sony a year early (the only explaination considering MS had by far the best games out in 2005).

Transitions frequently occur with the advancement in silicon processes, with an eye to quick downscaling to become affordable. The general outlook is as follows for the roadmaps:

2005 90nm
2007 65nm
2009 45nm
2011 32nm
2013 22nm
2015 16nm

And while some people shoot ahead (Intel) there is always the fact certain industries and mainstream consumer devices from everyone else seem to lag behind about 1 year. e.g. Intel already has 65nm processors out, yet GPUs *just* reached the 90nm process close to 2 years later (65nm was accellerated by Intel due to Prescott issues as well).

The technology spurning process changes has improved, but also the hurdles ahead are getting worse, so the 2 year average for a mass production silicon seems about right.

Ok, so why bring this up?

If MS aims for 2010 they may get stuck on the 45nm process which allows about 2x the transistors of a 90nm chip of similar die space. 400mm fabs are still a while off with most fabs just now switching to 300mm wafers as the standard so there wont be much price relief until then.

Process reduction is the #1 reason for performance increases because they (1) allow more transistors on a given die space and (2) allow the frequency to be increased. With all the leak and soft ground issues it is hard to measure how frequencies will improve (indeed, even Intel as gone back to the wide and slow approach instead of the narrow and fast).

So we are left looking at the process the chips are on. Lets look at the GPU since it is the easiest to look at scaling. A die with 2x the transistors would only yield 2-3x the performance (if we assume that there are no new significant needs like a large geometry shader or whatnot and that processing logic continues to be small; e.g. shader ALUs on the X1000 series are only about 2M transistors, so moving from a 16 ALU design to the 48 ALU design was relatively cheap).

2-3x the performance (plus whatever memory and frequency improvements) is not enough to merit a "next gen".

The problem becomes more noticable on the CPU side as they do not scale well in regards to parallelism at this point. Adding more cores is not enough to get more performance. Sure, if MS decided to go with 3 out-of-order execution PPC CPUs with a ton more cache and at ~5-6GHz they would get a boost, but not the typical 8-16x increase typically expected in a gen-to-gen jump.

I have no doubt Sony will be probably targetting 2012. That will leave the PS3 on the market for 6 years and allow BR to settle in--and for Sony get get its followup read. This will allow them to play hopscotch: PS1&CD, PS2&DVD, PS3&BDR, PS4&_?_.

If MS went for the 45nm process and Sony hit the 32nm process MS would be in deep trouble. The first reason being the obvious: Sony could either cut costs and launch with a lot more for less, or they could match costs and have a system easily 50% faster--and maybe more with solid design. The second issue is CELL is aimed to scale. The original goal was more than 1 CELL in the PS3, alas, the original targer was ALSO 65nm. CELL's FlexIO is designed for chip-to-chip communication (with over 100GBps of bandwidth). If Sony is able to get ahead on process technology by waiting a year Sony will get to ride out the PS3 (something they would love to do... milk the install base!!) and then offer a console with more CELLs in it.

The Sony strategy with CELL is very forward looking. I personally don't think CELL is the best thing for now, but as a visio for a *platform* and *scalability* Sony really has positioned the PS3 as forward looking. CELL *does* scale, much better than the chips the 360 is using, and really bodes in Sony's favor.

The final reason I think MS will go for 2011 head to head with Sony is because MS's 360 is more profitable this time (it will reduce in cost very nicely) and MS wants to regain money. The best way to do this is to continue launching games and NOT launching a new console!! New consoles = kill old console game sales.

We know Sony and Nintendo wont launch in 2010 because that would be a 4 year cycle; so 2011 looks the soonest. So by going with 2011 MS can regain some profitability by their early leap in 2005 and also allow devs the time to work on their new games longer (since everyone will leap at 2011/2012). Even better for MS is if Sony went an extra year again and MS would still have a year head start AND get to milk the 360 for 6 years.

Anyhow, just my thoughts on that. I could be wrong... but I am betting on 2011 for MS ;)