X-Box 2 Speculation!

Deepak

B3D Yoddha
Veteran
We havent heard anything from MS about XB2 so I wonder what they are thinking...

BTW, MS will try its best to make sure that XB2 is as powerful as PS3 if not more....about output/triangles/TFLOPs/bandwidth, I am not an expert so maybe you guys could help...

I see following happen....

1) Becse it is to late now to design a custom designed CPU/GPU for XB2, MS will follow tried and tested formula.....means CPU from Intel or AMD....better if they opt for AMD, maybe the new 64 bit...Sledgehammer or whatever....

2) GPU from nVIDIA as ATI has already sealed the deal with Nin...and MS cant take chance with any other player, better for backward compatibility also...

So we are not going to see anything surprises as far as XB2 is concerned...

speculated XB2 specs....

5GH 64 bit Hammer CPU
NV50 nVIDIA GPU
DDR3 RAM
nForce4/5 mobo
BlueRay DVD-ROM
Hard Disk
Ethernet/ATM adapter

etc
 
err mabye a 3-4ghz hammer don't think any faster

mabye an nv50 i think they should go power vr though


what the hell is ddr 3 ? mabye qdr or whatever that ram is ati is going to be using soon .

don't see them using a nforce board it will most likely be custom for the best bandwidth on everything , if using a hammer chip the nortbridge will be on die basicly .

I dunno if they will push blue ray though mabye a normal dvd .
 
I am sure MS will stick with nVIDIA....for they would not like want to try anything exotic....then backward compability is a must...

so safer to go with NV than PVR....

latest NV GPUs are using DDR2...so by 2005 we may see DDR3....
 
Deepak said:
I am sure MS will stick with nVIDIA....for they would not like want to try anything exotic....then backward compability is a must...

so safer to go with NV than PVR....

latest NV GPUs are using DDR2...so by 2005 we may see DDR3....


And who's to say that Nvidia would go with Microsoft again?? :LOL:
it's a 2way relationship, it's not like MS decides and NV submits.... :?
 
Deepak said:
I am sure MS will stick with nVIDIA....for they would not like want to try anything exotic....then backward compability is a must...

so safer to go with NV than PVR....

latest NV GPUs are using DDR2...so by 2005 we may see DDR3....

but ddr 3 is a myth i see it on no road maps and i've barely seen any ddr 2 . I see qddr ram or whatever the ram that ati is going to use in future graphics cards but no ddr 3 on any road maps . Considering that in 1999 ddr first came to video cards and its not 2003 and the first cards are coming with ddr 2 .... i don't think we'd see it by 2006 at the earliest and thats only if they can't up the mhz anymore . If they make a new dram it will most likely be a quad ram not another dual ram after this .
 
Rambus Yellowstone (or was that Redwood? hehe, too darn many colors!), is already a quad data-rate signalling memory technology. Now all we gotta do is actually implement it in silicon, hehe. :D

However, it seems bandwidth isn't the biggest concern anymore these days, it's latency. The crappy old DRAM cell is holding us back far more than anything else, when a modern microprocessor can stall for half a thousand clock cycles due to a memory page miss you know something's basically WRONG with your design.

Schemes in the past have tried to cut down on latency with SRAM caches/buffers etc (like 1T SRAM and others), but so far people have been too concerned with cost. A bigger memory die means more expensive memory, but DRAM is dirt cheap anyway these days and with bigger and bigger wafers and higher integration than ever before (wasn't Nanya going to build a 12" wafer plant? Man, that's a BIG silicon disc!) should mean that plays much less of a role than in the past.

Hopefully we'll see something like 1T SRAM coupled to a high-bandwidth interface used in main memory within a reasonable timeframe. If I was Bill Gates, I'd DECREE something like that had to happen for windows longhorn certification or something like that, heheh. :D

*G*
 
I dunno i think the main problem is how out of sync the ram and cpus are . We have 3 ghz chips using 400mhz ram on a 200mhz bus (800mhz considering the p4 is quad pumped ) I wouldn't mind seeing how a 3ghz chip would performe with 3ghs ram on a 3ghz bus .
 
I don't think going with PowerVR is risky. I mean obviously if MS didn't think they currently had the technology they needed then they wouldn't go with them. That goes without saying. But otherwise I see no problem with them bringing a chip to market on time for XBox2. MS's financial backing would ensure that.

Also one important thing to consider is, unlike Nvidia, PowerVR are happy to allow MS themselves to produce the chips. They only want licencing fee's. So that would mean MS would have total control over the chips. They could decide when to go to a smaller process and save money. There would be no need for court cases.
 
Here is some informed speculation on what MS may be thinking about Xbox 2, and why we haven't heard anything about it yet. From an interview with Nvidia's PR VP, Mike Hara:

http://www.penstarsys.com/Interviews/nvidia/m_hara/index.html

There are no "official" plans for an X-Box 2, but Microsoft is of course continually re-assessing the situation. As such there are no concrete plans from any manufacturing concerning X-Box 2. When information does come out, it will be from Microsoft and not a 3rd party like NVIDIA or ATI. In other words, X-Box 2 is more rumor than reality now. Sony is very much in the lead with the PS-2, and they have introduced very early specs for the PS-3 which is slated for a 2005 introduction. These specs show a very capable product utilizing the Supercell processor, but with such a specification and open ended architecture, Sony could easily raise the bar when competing products are announced. Microsoft has a very daunting task to produce something to compete, and Microsoft does not want to expose its hand until Sony is firmly committed to the design and specs of the PS-3.

Microsoft is taking its own sweet time in choosing a partner, and doing so at an early date would only hurt its chances at producing a competitive product. This also keeps AMD and Intel in the running as a future partner, as both have very promising future products to base the CPU off of. The big question for Microsoft is exactly how much CPU power the X-Box 2 will need? The current X-Box has a 700 MHz Pentium !!! derivative, and that is not exactly a vector processing powerhouse. The PS-3 is all about vector processing, but seemingly not so much about graphics, so this could be an area where Microsoft may have an edge. We will not hear anything from Microsoft this year, and of course announcing anything in 2005 would be too late. From this we can infer that Microsoft will probably make an X-Box 2 announcement in early 2004. Another reason for Microsoft not wanting to announce X-Box 2 early is that it wants to keep excitement about the current X-Box at a fever pitch, and not dilute it by announcing specs on future products. NVIDIA has worked very closely with Microsoft, and they fully expect to work with Microsoft again.

The current X-Box still has a lot of headroom left in it, and future titles will help to expose that headroom. The Halo series will again come to the fore in the next few months with the introduction of Halo 2. Now that developers are comfortable with programming for the X-Box, more graphically intense titles will be hitting the shelves very shortly. Halo 2 will feature some very impressive graphics by uncovering the advanced bump mapping capabilities of the X-Box GPU, as well as increased lighting complexity. The next generation of games will keep enthusiasts quite happy with the X-Box. This is a very natural progression for consoles, as the initial games are typically not as impressive graphically as those that come several years after. We have seen this with the PS-1 and now the PS-2, and the X-Box will be no different. Developers will find ways to introduce advanced rendering functions that were initially absent, and this helps to give that particular console a longer life as well as greater perceived value.
 
fbg1 said:
The PS-3 is all about vector processing, but seemingly not so much about graphics, so this could be an area where Microsoft may have an edge.


well, having a vector processing monster would be good for a little thing called Animation, which is where i can see the most room for improvement in the next generation... i rather see a very high detailed model move naturally than a super-high detailed model moving like in the current generation... i rather see water moving like real water than super textured bumpmapped stuff moving like the water in the current generation...
see what i mean...
 
Deepak said:
1) Becse it is to late now to design a custom designed CPU/GPU for XB2, MS will follow tried and tested formula.....means CPU from Intel or AMD....better if they opt for AMD, maybe the new 64 bit...Sledgehammer or whatever....


Its not too late if they were designing it since Xbox1 was finished? I don't know if MS has the ability to pull it off. But if designing its own CPU/GPU was an option, wouldn't that be the cheapest route for them.

Also, will MS want to just pull the fastest CPU from AMD or Intel now that their respective top end processors are generating 75-100watts of power. Thats alot of heat to remove from even Xbox's massive frame. Now I'm guessing everyone and their mother will be on 65nm by 2005. Whats Intel's roadmap (10 GHz by then?). I assumed a more senseable route would be to use Intel's Banias core, that thing should scale well and keep heat to a minimum , if not the fastest chip on the block. Furthermore, aren't GPUs like NV30 dissipating a ridiculous amount of heat now as well. I suppose they can always downclock it, but what are the odds of Xbox2 being water-cooled :LOL:

Lastly, would MS go with a Athlon64/Opteron/Itanium. What are the practical advantages of a 64-bit CPU. More than 4GB address space? Will next gen consoles have more than 4GB of memory :oops:
 
london-boy said:
well, having a vector processing monster would be good for a little thing called Animation, which is where i can see the most room for improvement in the next generation... i rather see a very high detailed model move naturally than a super-high detailed model moving like in the current generation... i rather see water moving like real water than super textured bumpmapped stuff moving like the water in the current generation...
see what i mean...

I noticed that too. Good point. Additionally, with enough FP power, they can do all the desired effects in software anyway. Early rumors have put the PS3 Cell at 8 VUs. Some of those ought to be available for bumpmapping, lighting, etc. Sony has always preferred raw programmability over hardwired gfx effects anyway.
 
fbg1 said:
I noticed that too. Good point. Additionally, with enough FP power, they can do all the desired effects in software anyway. Early rumors have put the PS3 Cell at 8 VUs. Some of those ought to be available for bumpmapping, lighting, etc. Sony has always preferred raw programmability over hardwired gfx effects anyway.


thanx, well whatever happens, i'm pretty sure that in the next generation of hardware Pixel Shader everywhere will be a given even in next-gen Army Men games... now that i think about it, we haven't seen an Army Men game in a while.... mmmmm... did anyone say anything to 3DO?? :LOL:

seriously, we are already seeing in XBOX games a decent use of bump mapping, it's pretty safe to say that PS3 games to use more pixel effects than a past-generation hardware...
 
a4164 said:
Its not too late if they were designing it since Xbox1 was finished? I don't know if MS has the ability to pull it off. But if designing its own CPU/GPU was an option, wouldn't that be the cheapest route for them.

I think MS could certainly design a chip. They pretty much dictate GPU design now with DX anyway. And their years of making OS's ought to give them some knowledge of CPU design. And they could always buy a small design house, like ArtX, to do the work for them.

Their problem is, can they design a CPU/GPU that will be competitive with the Cell? Sony, IBM, and Toshiba have spent a lot of time and $$ on this thing, and its beginning to sound promising. MS certainly has their work cut out for them.

I assumed a more senseable route would be to use Intel's Banias core, that thing should scale well and keep heat to a minimum , if not the fastest chip on the block.

I don't think MS can afford to use anything less than the best performing parts, in the face of PS3. If heat is a problem, they'll have to figure out how to cool it.

Furthermore, aren't GPUs like NV30 dissipating a ridiculous amount of heat now as well. I suppose they can always downclock it, but what are the odds of Xbox2 being water-cooled :LOL:

NV30's heat problems are due to the fact that it was designed for a .13 low-k dielectric manufacturing process. TSMC couldn't get the low-k part working, hence GFFX's heat problems and Nvidia's new manufacturing deal with IBM (who does have working .13 low-k). NV30's current heat problems aren't a good indicator of future NVxx heat problems.

Lastly, would MS go with a Athlon64/Opteron/Itanium. What are the practical advantages of a 64-bit CPU. More than 4GB address space? Will next gen consoles have more than 4GB of memory :oops:

Keep in mind that consoles have to meet that $299 price point (or perhaps $349 by 2005). I doubt for that reason they will have so much memory (especially whatever high-performance QDDR or Rambus variant is the rage in two years). Also, I seriously doubt Xbox2 will use Itanium2 either, unless it undergoes a significant price drop. Afaik, lowest-end Itanium 2's are still selling for >$1000.00. Athlon 64 or Pentium 4/Pentium 5 are the most likely candidates. The 64bit advantage Athlon 64 provides is better performance from increased registers, and better FP performance and precision (which Xbox2 will need given PS3's high FP numbers).
 
I remember reading an interview of Ken Kutaragi in which he explained the basic strategy behind PS3. Sony's intent was to forecast the state of PC technology in 2005, and to design a PS3 system more advanced than that PC technology. The thinking was, MS is a PC company, Xbox is based on the PC ecosystem, therefore Xbox 2 will most likely be based on the that ecosystem. If Sony could leapfrog that ecosystem with innovative, out-of-the-box, non-legacy technology, then they could eradicate Xbox's only selling point over PS2 - its better technology.

It's a bold strategy, and seems to be coming along nicely with IBM's help. Mike Hara's statements sure make it sound as if MS is in a most difficult fix now. I'll look forward to seeing what they can come up with for Xbox2.
 
If MS planned to lose money, they might as well give away games for free than spending money developing hardware to beat PS3. Xbox2 launched with 10 free games at $300. That would be hard to beat.

Average consumers don't really care if MS or Sony lose money on hardware, but if they see Xbox 2 with 10 free games for $300, they know its good value :)

MS should also price wars on software front. Cheaper software for Xbox, would also lead to more Xbox sales.

I think that's how MS should lose money, if they wanted too. Once they force everyone out of business, than they can have monopoly and bring back the cash.
 
a4164 said:
Its not too late if they were designing it since Xbox1 was finished?

Where would they get this world-class semiconductor design team from? M$ has never done anything even remotely similar in the past. They'd need like a couple dozen well-experienced engineers. We haven't heard anything in the rumor mill signalling they've gone down this path.

No, you fanbois gotta lay off that crack pipe, hehe. :LOL: M$ is going to use off-the-shelf tech again just like last time. I know, I know, you want your sugar daddy Billy Goat to whip out something even longer and phatter than Evil Sony's Cell, but face it, it's not going to happen.

But if designing its own CPU/GPU was an option, wouldn't that be the cheapest route for them.

What basis do you have to support such a standpoint? They'd create a team out of nothing, with zero experience to begin with other than what the people they hire bring with them, whip them into a world-beating design studio and come up with a monster CPU all in the space of a few years and all in total secrecy? Frankly, I don't see that ever happening.

Also, will MS want to just pull the fastest CPU from AMD or Intel

Of course they won't. What was the fastest out there at the time when M$ chose - GASP - a celeron 733 to power its superduper games console? Better yet, what was the fastest when the console actually BECAME AVAILABLE?

The fastest chips are monster expensive, even though the markup on CPUs is enormous it would still be prohibitively expensive.

Now I'm guessing everyone and their mother will be on 65nm by 2005. Whats Intel's roadmap (10 GHz by then?).

If they're pushing 6GHz by then I guess we should count ourselves lucky. You gotta remember, the P4 debuted at 1.5GHz. From there it's only little more than doubled in the space of several years. It's totally unreasonable to believe it would reach 10 in another two and a half years or so.

I assumed a more senseable route would be to use Intel's Banias core, that thing should scale well and keep heat to a minimum

Banias still draws 30+ watts at top speed, and most likely it will stay that way or get worse as speed increases. Besides, it's based on the aging P6 architecture which debuted wayyyyy back in 1995. Hardly worthy of a console set to debut a whole decade later!

but what are the odds of Xbox2 being water-cooled :LOL:

Extremely small. A well-designed heatsink can deal with 100+ watts of power dissipation no problem, though I bet it wouldn't be ideal from M$'s point of view. Biggest concern would be getting it quiet, though considering the huge box they used first time round, they could fit some giant fans in there and have them run at a low RPM...

Lastly, would MS go with a Athlon64/Opteron/Itanium. What are the practical advantages of a 64-bit CPU.

Hm, difficult to say. Itanium is out of the picture because of its enormously crappy price/performance (and its power consumption is equally monstrous to boot).

They couldn't do a uniformed memory setup with Hammer, because the Hammer's on-die memory controller would give high latency for the graphics processor. On the other hand, I'm not sure they'd particulary WANT UMA for the next round anyway, as it is basically just a cost-saving measure that limits performance of all components in the system.

I'm pretty sure Inhell will win the contract again with a fairly high-GHz hyperthreaded P4, probably using a 400MHz bus (real MHz, I mean). Getting the P4 to GHz higher than the PS3s Cell will be vital in the marketing strategy to make it sound faster to the masses, despite it getting the proverbial offal kicked out of it on a pure numbercrunching level.


*G*
 
Back
Top