Couple more things:

All new consoles are going to need to support HDTV. That means resolutions of 1080i (1920x1080) and 720p (1280x720). So the GPU and CPU processing power is definetly going to need to be a lot more powerful. "Saist" mentioned Itanium, I find this highly unlikely. Itanium is way to expensive. Sony has to delay PS3 and Cell technology will not be on time (plus it will be really expensive). This is why you are seeing PS2.5 and PSX coming to market.

I still think the CPU will be 32-bit. No need for 64-bit (not yet anyway) and it is too expensive. Lets not forget that these things need to sell at $300 (maybe $400-500, but that is pushing it). John Carmack said that 64-bit won't make that big of a diference in games (Doom III at least) so there is really no point. It is not like we are going to see game consoles using over 4GB of ram (I think that is two generations away).

AMD64 and Epic Architectures are more powerful than x86, but I think x86 can still handel it. Xbox uses a 900MHz P3 basically (I think), so going to a 2.2GHz AthlonXP or a 3.0GHz P4 should give it all the power it needs. More important is the GPU ( and memory bandwidth).

try again: X-box uses a 733mhz Celeron processor. It's about 10% faster than a stock off the shelf Celeron. But it's still a celeron.

There also is the problem that CELL's current archicture is far above X64. Cell prototypes are showing a power-level of processing equivelent above Pentium4's chilled to allow 4ghz operation. Far Far above. As in the beyond therotical 5-6ghz range of the Pentium designs. In other words, a 2.2ghz AXp or 3ghz P4 isn't going to cut it fighting with CELL.

Also: little fact check. Consoles are already at 64 bits and 128 bits. The GCN's PowerPC while reported as being based off of G3, is actually based off IBM's POWER4 line, and is a slow cousin to the upcoming PPC 970 (marketed as G5). Process per process the GCN's little PPC chip dominates the X-box Celeron, and it also has the capability to process 2 different 32bit threads at the same time, regardless of thread type (TRUE HYPER-THREADING). The PS2 chip is a full blown 128bit core, and it's 2 vector engines can offload and process 99% of all processes written for the 128bit core. Again, processing wise, the X-box falls very far behind.

Once you start talking the power level's of the processors that will be used in the PS3 and Nitnendo's Game-Next... it becomes pretty clear. X86 won't be able to handle it.

geez, need an edit button on this thing.

There also is the problem that CELL's current archicture is far above X64

supposed to be

There also is the problem that CELL's current archicture is far above X86

Saist, I want a high powered console as much as the next guy, but what you did not mention at all was the cost. These consoles are supposed to go for around $300. Even if they were packed with all the technolgy posible I don't think you could sell them for more than $500, because your market would shrink too much. M$ was already loosing over $100 on each Xbox, I don't think they can afford to loose $300-400 on each Xbox. Don't forget Micro$oft's bottom line is profits.

Sony PS3 rumored 2005 release.
Xbox2 rumored to be released before PS3.

Generally systems are released before Christmas so it will probobly be late 2005. That gives them aproximately 2.25 years to have a final product. But you also have to remember that games need to be developed for the new console. A certain amount of this can be done in the absence of Xbox2 but at some point they will need it. I think they need to have the hardware done in 1.25 years, giving software a year.* So what you have to think about is what CPU is going to be cheep 2.25 years from now, when Xbox2 is being mass produced. It seems to me that Prescott/Athlon64 will be the cheepest. I think the choice will be Prescott, unless the memory Athlon64 memroy controller is changed.

Again M$ is focused on money, so to keep there development costs down I suspect that they are going to stay with x86. "it becomes pretty clear. X86 won't be able to handle it." You should ask Intel about x86 being dead, I suspect they would think otherwise.


*Speculation. I know that they do need to have there final hardware locked down at some point, but if someone knows more specifics, it would be apreciated.


I too wish there was an edit button. This debate has actually become quite interesting.

Indeed the cost is prohibitive. CELL is delayed at least until 2006 under the current budget.

And the cost factor is indeed paramount, all one needs to do is look at Trip Hawking and 3DO.

I think Microsoft is losing more than $135 per X-box sold... trying to get real numbers out of what they publish to the US IRS and what they actually consumers is difficult at best as the numbers rarely ever match. That's just an estimate made by somebody I know who's got their master's in accounting and after they spent a week attempting to match what Microsoft was telling the public about it's X-box figures, and what Microsoft was telling the US governemnt... and also... the numbers of sales weren't matching either... according to Microsoft's tax slips, it's X-box is in a decided 3rd place. Getting confirmation of that... well... is impossible.

The focus on money was also hammered recently by Ballmer,

(see post here: http://www.thatforum.com/showthread.php?s=&threadid=29 82 )

And I'm not talking about X86 being dead. I'm simply saying that with what IBM, Sony, Toshiba, ATi, and Nintendo are planning as is known right now, the X86 platform won't provide enough punch at it's cost level. Yes, the PS3 CELL will be only 4 or 8 PPC's linked together, so it won't be the prohibitive cost 16-32 PPC links we'll see for most servers. Providing CELL actually works out of a test lab.

Right now as it stands, and this is coming from having talked with several of the vendors invloved, X-box2 plans are on Haitus. And it also appears to be on haitus because investors aren't sure they want it to go foward (see ThatForum link above)

more as I have time: currently at work inside Cox Communications... yes, I'm tech support. yes, I know my life is hell. Yes, I know I have to go back out and take stupid calls.

But it pays good.

I guess there is one other thing to think about. Will there actually be a game engine that takes advantage of Sony's Cell technology. DoomIII will be the most graphically intense game to hit the market in a long time, it has been in the making for a long time.

With the addition of HDTV the high resolution will definetly require consoles to work a lot harder. But I think because of the lower frame rates that can be accoplished with out too much difficutly and with out excessivly expensive processors.

However, it would be cool to see PS3 with Cell processors and 1024-bit memory controller (thought I read that some where).

Toshiba announced... I don't know how long ago, that they had 1024-bit memory bus working with prototype YellowStone based RDRAM. (yes, that's Rambus D-Ram) I just surfed for the link to it and couldn't find it. I do recall them stating however that it was working at extremely LOW speeds (I recall somewhere in the neighborhood of 10mhz) Considering that all current signs are pointing to the use of YellowStone RDRAM for the current PS3 archicture, while Toshiba did not say the design was for the PS3, as the PS3 is the most lucrative project with YellowStone RDRAM contracts... well, I think I know where that bit came from.


As to game engines... I really don't know. A lot of the recent fru-ha-ha I've seen over the PS3 (and Nintendo-Game-Next for that matter) is developers asking for concrete Dev-kits so they know what they are getting... since CELL isn't exactly working at a mass production level right now, game developers aren't sure what they will see in the end.

It is my opinion that the PS3 will follow the PS2 and PSX launch methods. The first titles while making some use of the power, won't even touch what the console is actually capable of. We won't see with the PS3 (or Game-Next) another HALO type game that is pretty much firing on everything the system has to give.