Predicting the Xbox-2

Vince said:
Dual GPUs? Perhaps your right, I just find the multichip paradigm too hard to swallow. Especially after MS's current problem with manufacturing cost.

Infact, David Hufford from Microsoft just stated in an interview (the same one where he said that as of now VALVe has backing away from their XBox/HL2 committment and the game isn't comming to Xbox) that they're still loosing money on each XBox sold - which is odd following the Nintendo/Sony trend after almost 2 years since launch.

Going with off-the-shelf parts / contracts has its advantages (time to market namely), but its a bitch to lower costs, especially when your volume isnt high enough.

I'm sure MS has 'learned' from this generation, but I still think they'll do the off the shelf route for XB2. I figure that if they are in this for the 'long haul', next gen will be another round of hardware bleeding, but they'll get their shit together for Act 3 and do it all in house, with a ~5 year R&D cycle to boot. There's just no way that MS can come up with something competitive, in house, in like 2 years to counter Sony/IBM's PS3 investment.
 
zurich i really think that by going with power vr and liscenseing a chip they will save alot of moeny.
 
Jvd,

What I tried to convey is so few words is that going multichip is a horrible thought from an economic perspective. You introduce cost (eg. packaging, PCB) that doesn't scale with lithogrpahy advances, it's fixed. Thus, it's way to expensive and will lead to economies of scale issues that make the current Xbox pricing situation look good in comparason.

I also fear for XBox Next if nVidia isn't involved as you propose... but we won't get into that right now.
 
jvd said:
zurich i really think that by going with power vr and liscenseing a chip they will save alot of moeny.

But then they loose a powerful technology partner - NVIDIA. Also, whos to say that PowerVR is up to the task? I mean, efficient, yes, but big numbers to out-PR Sony with? No. Also, since this is the (Direct)Xbox, I think it'd be a huge boon to have an active DirectX IHV who contributes and works on the API. Not trying to knock PowerVR, but they really have a fraction of the clout and influence, not to mention R&D resources, that NVIDIA has. Doesn't hurt that a massive wack of development workstations are powered in someway by NVIDIA ;)

You're also forgetting that, despite the recent PR fuckup, NVIDIA has alot of fans out there. The Xbox gets alot of exposure from these sites, and they tie in well to PC gamers (which is a resource that MS would like to slowly drain). As an NVIDIA fan, I know that I would be more than a little miffed if MS choose someone else for XB2.
 
zurich said:
jvd said:
zurich i really think that by going with power vr and liscenseing a chip they will save alot of moeny.

But then they loose a powerful technology partner - NVIDIA. Also, whos to say that PowerVR is up to the task? I mean, efficient, yes, but big numbers to out-PR Sony with? No. Also, since this is the (Direct)Xbox, I think it'd be a huge boon to have an active DirectX IHV who contributes and works on the API. Not trying to knock PowerVR, but they really have a fraction of the clout and influence, not to mention R&D resources, that NVIDIA has. Doesn't hurt that a massive wack of development workstations are powered in someway by NVIDIA ;)

You're also forgetting that, despite the recent PR fuckup, NVIDIA has alot of fans out there. The Xbox gets alot of exposure from these sites, and they tie in well to PC gamers (which is a resource that MS would like to slowly drain). As an NVIDIA fan, I know that I would be more than a little miffed if MS choose someone else for XB2.

except that nvidia bent ms over a desk when it came to the price of the xbox gpu.

Going with power vr (which they are rumored to have a new card coming out soon) would give them huge numbers

say a dual 500mhz 8x1 kyro. thats 4000 for fillrate. Times that by the over draw of say 4 . thats a fill rate of 16,000. Then if they go dual gpu they an double that. Thats huge for fillrate. We also know that powervr can make a good consle graphics card. They can make a great arcade unit and they can make a great tnl unit.

If the rumors of nvidia making a sony chip you can bet ms will let them go.
 
except that nvidia bent ms over a desk when it came to the price of the xbox gpu.

Going with power vr (which they are rumored to have a new card coming out soon) would give them huge numbers

say a dual 500mhz 8x1 kyro. thats 4000 for fillrate. Times that by the over draw of say 4 . thats a fill rate of 16,000. Then if they go dual gpu they an double that. Thats huge for fillrate. We also know that powervr can make a good consle graphics card. They can make a great arcade unit and they can make a great tnl unit.

If the rumors of nvidia making a sony chip you can bet ms will let them go

That was like 5 years ago. GPUs have become an order of magnitude more complex since then.

Really, in the last year or two, PowerVR really haven't shown that they're up to the task of producing anything that could rival ATI/NVIDIA's latest and greatest.

Also, if MS went with PowerVR that'd effectively kill backwards compatibility :? Whereas if they used an NV 5x or something, it wouldn't be too much of a stretch to force aniso/FSAA/HDTV res to spruce up the old XB1 games, I suppose.
 
I don't see how it will mess up backwards compat. I mean a dx 10 card should be able to do everything a dx 7 card can do and do it much better.
 
jvd said:
I don't see how it will mess up backwards compat. I mean a dx 10 card should be able to do everything a dx 7 card can do and do it much better.

Cept DX goes alot closer to the metal on Xbox, and exposes alot of NV2x specific hardware (buffers, registers, stuff - not a dev here :p). DeanoC made a post a while back about the differences.. IIRC it was a more complex version of what I just said :p
 
zurich said:
jvd said:
I don't see how it will mess up backwards compat. I mean a dx 10 card should be able to do everything a dx 7 card can do and do it much better.

Cept DX goes alot closer to the metal on Xbox, and exposes alot of NV2x specific hardware (buffers, registers, stuff - not a dev here :p). DeanoC made a post a while back about the differences.. IIRC it was a more complex version of what I just said :p

I understand that but we are talking about a chip that could offer more than 10x the performance of said chip. Who knows though.
 
it's not strictly an perfomence issue, more that your asking engineers to write a 99.99% accurate emuilator on hardware with sizable differences.

hell even PS2 which forgoes the software emulation for the most part has trouble hitting that figue.
 
Is it me or is XBOX 2 falling massively behind Moore's Law?
733Mhz -> 5Ghz in 5 years is like 2x behind the curve.

Vince:
Does multi-core chips on the same die or package sovle the scalability problem?
 
JF_Aidan_Pryde said:
Is it me or is XBOX 2 falling massively behind Moore's Law?
733Mhz -> 5Ghz in 5 years is like 2x behind the curve.

Vince:
Does multi-core chips on the same die or package sovle the scalability problem?

i thought more's law was transistor counts not mhz .
 
i thought more's law was transistor counts not mhz .

thats not strictly true, its the equilibrium reached between the cost/density ration for any lithgrophy process that doubles per 18 monthes.

there a nice article at arstechnica on moores law and it's implications.
 
JF_Aidan_Pryde said:
Does multi-core chips on the same die or package sovle the scalability problem?

But of course! So, whats new with you bud? Haven't seen you around in... ohh, awhile.

Anyways, by moving them on-chip you're eating up your tranistor budget which does scale - and scales very well at that.

PS. Yes, it will be interesting to see if Xbox Next manges to preserve backwards compatability inlight of the fact that many developers are forgoing abstraction and coding directly to nVidia's PushBuffers - which I'm guessing would be architectually discinct.

Perhaps a developer could elaborate?
 
JF_Aidan_Pryde said:
Is it me or is XBOX 2 falling massively behind Moore's Law?
733Mhz -> 5Ghz in 5 years is like 2x behind the curve.

LOL, you realize of course that all the specs youre reading here are completely made up, right? for all we know it could be a quantum processor sent from mars running at 3 billion gigahertz....

XBox2
- Dual GPU configuration (one ATI and one Nvidia)
- 1,024 banks of 1,024 gigs of ram
- DirectX v.376
- CPU: 3 billion gigahertz processor sent by spacemen from the future
- has robotic legs so it can do a little dance when something really cool happens in a game
- just the controller alone has the processing power of 20 PS2s
 
notAFanB said:
thats not strictly true, its the equilibrium reached between the cost/density ration for any lithgrophy process that doubles per 18 monthes.

there a nice article at arstechnica on moores law and it's implications.

And its implications.

Not Moore's Law itself... which does, in fact, deal solely with transistor counts and "Performance" but not MHz value.
 
jvd said:
I'm not . If what i think is gonig to happen happens i will be happy. I epect about 10 x the performance of todays games with all the effects of dx10 used.
I'm expecting a big jump from this generation as well (better AI and physics is what interests me the most), but he's expecting too much.
 
Back
Top