MS wants XBox2 out before PS3?

[megadrive0088]

Perhaps NV50 is too far off for it to be apart of XBox2, however, there is no way XBox2 will be based on a NV3X chip. the whole NV3X line will be outdated by the time XBox2 is ready to go, and NV3X will be no where near good enough to be competitive with PS3 in 2005-2006. We are talking NV4X series in XBox2 at least. correct me if you think this is wrong here.

Look back to the first rumors of X-Box in 1999. At that time Nvidia had NV10 (Geforce256) out. It was thought at that time that X-Box would use an NV10 for graphics. a few months later, it was thought that XBox would use NV15. When XBox was officially annouced in March 2000, the graphics chip was going to be NV25. now things got a little confused and it turned out that XBox would use a modifed NV20/GeForce3, called the NV2A which turned out to be inbetween NV20 and NV25.

NV30 is 2002 technology launching in late 2002 or early 2003. I highly doubt an "NV3A" would be used in XBox2, even if NV3A was a little bit beyond NV35. At the very least, XBox2 is going to be DX10 and NV40, it well could be DX11 and NV50, but certainly not DX9 and NV30 or NV35.

unless MS was launching XBox2 in late 2003/early 2004 which is extremely doubtful.

 

[megadrive0088]

NV30's delay does not necessarily have any impact on NV40's schedule, independent teams and all, even if they are shooting for a 2 year cycle on average nowadays NV40 could be ready in time.

well said, and I agree.

NV40 would be the absolute minimum that I would expect XBox2 to be based on. IMHO. I believe that NV40 will be the starting point for XBox2 development since the NV4X series is the most likely architechure for XBox2.

 

[megadrive0088]

keep in mind, that XBox2 is 3-4 years away. NV3X will be old, old technology by then. whatever XBox2 uses for graphics, the actual XBox2 version of this chip has not yet been designed, nor its base architecture.

 

[Vince]

We are talking NV4X series in XBox2 at least. correct me if you think this is wrong here.

Ok, I think this feeds into it:

NV30 is 2002 technology launching in late 2002 or early 2003

We don't know what the Nv30 implimentation of CineFx is capable of. While we know much about CineFx, we don't know the architectural details of the NV30 and I, for one, feel that we're going to be in for a surprise. There is much that can be done to increase the pathetic effeciency of current IMRs (which totally embody the "bruteforce" ideal) and that such schemes - which would allow for the computational effeciency that nVidia spoke of - would allow developers to actually utilize complex shaders to a high degree. Something that your not seeing with present solutions, something that would take a normal IMR several generations to achieve with the rudimentary and strictly evolutionary progression we're currently seeing from the likes of ATi....

But this is just my gut feeling - although it's worked out well in the past.

Look back to the first rumors of X-Box in 1999. At that time Nvidia had NV10 (Geforce256) out. It was thought at that time that X-Box would use an NV10 for graphics.

Actually, anyone read the Opening the XBox book? IIRC, the rumor had validity as MS was internally debating the form that XBox was to take...

It was thought that XBox would use NV15. When XBox was officially annouced in March 2000, the graphics chip was going to be NV25. now things got a little confused and it turned out that XBox would use a modifed NV20/GeForce3, called the NV2A which turned out to be inbetween NV20 and NV25.

Yeah, I know this I think you've told the tale of the evolution of MS's vision of the XBox as the project progressed - not that people were just lying and totally off. I think theirs alotta validity in the above comments, but in the way that the rumors were based on factual information that evolved with the 'box.

At the very least, XBox2 is going to be DX10 and NV40, it well could be DX11 and NV50, but certainly not DX9 and NV30 or NV35.

Whats this infatuation with big numbers? You do realise that nomenclature mean nothing, right? You don't know anything concrete about the NV3x implimentation of CineFx, and you sure as hell don't know of their timetable - but we do know their contemporary history and the time it takes to go from a product specification (Which MS just sent out) to implimenting these in an architecture and getting a completed netlist - then to finish up the back-end layout of the design and get an GDS-II tape from the netlist. We also know product ramping times for XBox (Anyone rememeber?). So, stop throwing out the biggest number possible and do the reverse math.

DirectX? Do we even know if the XBox will use a conventional 'API' such as this? I'm interested to see how high the level of extraction becomes in time - I've heard one developers comment on programming to an OS

 

[CaptainHowdy]

If MS did this, being its 2005, that would be shooting themselves in the foot, giving thier console less than a 4 year life..

Microsofts target is not to gain money with Xbox 1. The cost for the hardware and marketing is too high to pay it back with software royalities. The target is to beat SONY and Nintendo in the long term. This is a very long term strategy (~10 years). They will put money in that market until there is no real competition left. Then they can start making huge amounts of money.

think you missed my point, my point was, noone will adopt a new console so soon after its previous generation.. Xbox doesnt have a strong enough user base to take such chances, it needs to wait and keep its support on the first one long enough that every owner feels satisfied.
they were told for years that it would last well beyond the other consoles because of how much more powerful it was, thinking your going to get a good 6-8 years of gaming out of it, sure it was a hella sweet deal, even at $299..but to have spent that, to find a new console is on its way so soon, I would be mad.

 

[Vince]

Actually, IMHO, your best bet would be to figure on a December 2005 launch, then using XBox figures as a base - trace the time back threw stockpiling and manufacturing ramp-up and then use an average of 3-4 respins during the back-end phase and get to a date [roughly] when the completed netlist is delivered. Then look at what Lithography processes and libraries are availiable or projected for that time and go from there with whats possible and not - both temporally and physically [on the die].

keep in mind, that XBox2 is 3-4 years away. NV3X will be old, old technology by then. whatever XBox2 uses for graphics, the actual XBox2 version of this chip has not yet been designed, nor its base architecture.

You stated a 2005 launch. It's basically 2003. Thats 2 years of solid development or research leading upto it. 3-4 years away is pretty ignorant considering they have to lock the specs and manufacture/stock-pile the thing.

 

[megadrive0088]

yeah but look, NV30 is most likely completed now, and its ramping up right? you don't seriously believe that XBox2, a console that is 3-4 years away from release, and probably 2 years from being completed would use 2002 technology, or do you?

Now, I never said that Xbox2 would not use CineFx. CineFx might span several generations of Nvidia GPUs --- and Nvidia works on several GPUs at one time. NV30 is done now, most likely. NV40 is being designed as we type, and NV50 is at least a twinkle in Nvidia's eye, at least on paper some of its specs are down. I would bet you ANYTHING that XBox2 uses some varient of NV40 at least, not NV30.


Unless XBox2 is launching next year, which it is not, it would not be using NV30 or anything in the NV3X line. IMHO, saying today that XBox2 would use NV30 or NV35 is like being in 1999 and saying XBox will use NV10 or NV15. No one really knows what XBox2 will use other than the fact that it will not use current technology NV30 is current technology as far as Nvidia is concerned, and its about to be current technology for the consumer, in a few months. Xbox2 is simply too far away for NV30 to be used. but as I said, possible not CineFx. XBox2 in 2004 with NV3X... that kind of talk also reminds me of a year or so ago when some people and websites thought that Sony would have PS3 out by 2003 using EE2 and GS2. I will leave it at that.

 

[V3]

As for Japan, MS need to change its strategy, that is all. Japanese goes with the fashion, if you can set the trend, it will sell.

You don't need good games, or anything, basically set the trend. That's all there is to it.

 

[BenSkywalker]

You stated a 2005 launch. It's basically 2003. Thats 2 years of solid development or research leading upto it. 3-4 years away is pretty ignorant considering they have to lock the specs and manufacture/stock-pile the thing.

Remember that the NV2A was several months ahead of its PC counterpart. If the XB2 were to launch today it is likely it would ship with NV30(MS would find a fab that could pull it off costs be damned, I think we have seen that enough). If need be, I could see them approaching Intel or AMD and using their fab capabilities(Dresden has plenty of room ATM) tied in to landing the CPU for the XB2.

Looking at late '05 I think that NV5X is not unlikely, nor is a DX11 implementation. It would serve MS's interest to have the next generation of XBox launch with a DX revision that exceeds what is available on the PC at the time(at least in a public fashion). Given that nV has completely seperate dev teams for their generations and the hold up with the NV30 has been the fab capabilities it is likely that NV40 is less then eighteen months away right now(although they may push it back depending on market conditions). NV50 ready in three years for the XB2 launch I see as more likely then not. It may be pushing it, but it is possible that we could see an early NV6X part by the end of '05 for the XB2 launch(given how late the NV3X is it is possible that NV4X is less then a year off).

 

[megadrive0088]

Remember that the NV2A was several months ahead of its PC counterpart. If the XB2 were to launch today it is likely it would ship with NV30(MS would find a fab that could pull it off costs be damned, I think we have seen that enough). If need be, I could see them approaching Intel or AMD and using their fab capabilities(Dresden has plenty of room ATM) tied in to landing the CPU for the XB2.

Looking at late '05 I think that NV5X is not unlikely, nor is a DX11 implementation. It would serve MS's interest to have the next generation of XBox launch with a DX revision that exceeds what is available on the PC at the time(at least in a public fashion). Given that nV has completely seperate dev teams for their generations and the hold up with the NV30 has been the fab capabilities it is likely that NV40 is less then eighteen months away right now(although they may push it back depending on market conditions). NV50 ready in three years for the XB2 launch I see as more likely then not. It may be pushing it, but it is possible that we could see an early NV6X part by the end of '05 for the XB2 launch(given how late the NV3X is it is possible that NV4X is less then a year off).

Ahhh THANK YOU. Finally, someone who is understanding what I am saying, and that has a reply that makes total and complete sense.
Now, I am not saying that XBox2 will use NV6X or NV5X, but I can say with near certainty that XBox2 will not use NV3X. My honest best guess is NV4+ (meaning between NV40 and NV45) at the least and NV55 at the most. A late 2005 release seems most likely at this point, with reports going around that Sony wants to keep PS2 around as long as possible, not launching PS3 until 2006 (which is not late, according to their 1999 plan) XBox2 in late 2005 seems perfect timing, giving the XBox a full 4 years before introducing a more powerful machine.

 

[megadrive0088]

of course, a similar outlook could be estimated if MS chooses to go with ATI. An ATI Radeon powered XBox2 would obviously use no less than a R500. It's well known that both R400 and R500 are being worked on simultaneously. Perhaps R600 would be ready in time for XBox2 since R600 will likely be ahead of NV60 in timeframe. If R400 is a 2003 product, R500 a 2004 product, certainly R500 is the least we could expect from an ATI based XBox2. R600 is probably pushing it, but could be ready in 2005 especially if ATI is really speeding up its chip development like they appear to be doing. ATI would be more than willing to speed up development of an chip that might just be able to make XBox2's timetable. ATI would like nothing more than to score the XBox2 contract, specially if Nintendo drops out of the console race.

Sorry to tell history over again but I wanted to make this point: You can also look at the MS/GigaPixel situation back in 2000. GigaPixel had only prototypes of its GP1 chip. yet it was going to try to deliver a GP4 chip for XBox--A chip several generations beyond what they had silicon of (GP1) So we really cannot know what XBox2 will use, all we can know for certain is that it will not use what is about to be current technology.

 

[aaaaa00]

You stated a 2005 launch. It's basically 2003. Thats 2 years of solid development or research leading upto it. 3-4 years away is pretty ignorant considering they have to lock the specs and manufacture/stock-pile the thing.

"Inside the Xbox" states basically that entire project was completed in about 20 months.

The initial idea for Microsoft to build a console was proposed in mid-1999.

The second half of 1999 was research, feasability studies, and "wouldn't it be cool if..." discussions.

February 2000 was when the final approval for the project came down and real design started happening.

March 2000 was when the project was announced at GDC.

February 2001 was when the NV2A and MCPX were finalized and taped out.

September 2001 was when manufacturing started for launch.

November 2001 was the launch.

All said, 18 months from approval to manufacturing, 20 months in total from approval to retail.

The XGPU is essentially an NV25 (more or less). The PC version (GeForce 4 Ti) launched aproximately Feb 2002, or about 4 months after the xbox.

So given NVidia's patterns (assuming NVidia maintains their 6 month cycle, and NV30 launches in January), we can probably expect them to be at something resembling an NV50 or NV55 by the time we hit the end of 2005, which is when any console launch for 2005 would be planned.

Presuming xbox2 follows the pattern set by xbox1, initial studies will occur in the second half of 2003, design work will start in early 2004 with lockdown in early 2005, and the final product will release in late fall 2005 with something resembling an NV55.

Note this is all highly speculative (ie. pulled out of my ass ) and relies on many things which could change at any time, like NVidia's release schedule and chip numbering schema, or ATI being chosen as the supplier, or even Microsoft deciding to go with a custom architecture and designing their own chips.

 

[Vince]

Looking at late '05 I think that NV5X is not unlikely, nor is a DX11 implementation....

It may be pushing it, but it is possible that we could see an early NV6X part by the end of '05 for the XB2 launch(given how late the NV3X is it is possible that NV4X is less then a year off).

Woah boy... NV6x in 3 years? I guess this comes down to a fundimental question of how nVidia's timetable will stay. I see a slower paced enviroment, with the pace set by less frequent, but more radical DX revisions.

I don't see the continued investment of producing entirely new cores or major revisions every year. CineFX is here for awhile - just as the TNT core lasted all this time as the foundation of nVidia's products.

Given that nV has completely seperate dev teams for their generations and the hold up with the NV30 has been the fab capabilities it is likely that NV40 is less then eighteen months away right now

What? Um, I smell something odd....

nVidia conference call, no Nv30 tape out => August

http://www.beyond3d.com/forum/viewtopic.php?t=2112

NV30 just completed INITIAL tape-out (this week)
-First OEM samples end of October
-Foundry is TSMC, not UMC
-There were several things that had to be dropped in the design when they realized they were going to be very late with the part. This decision was made several months ago. The most notable feature was the primitive processor. This will show up in NV35.

http://www.beyond3d.com/forum/viewtopic.php?t=2385&start=0

AFAIK, the CEO (whose name eludes me) has stated several times when pressed that no tape-out occured untill early September. The problem - while the Fab's early problems add to it - probobly stem from the influx of ideas injected into the NV30 project when the 3dfx aquisition took place. The design was probobly too ambitious for the temporal and lithography contraints. This, together with the lithography production problems caused the problem their in now.

But, keep in mind why it took to long to tape-out. I mean, allowing 1 spin for production quality silicon on a new architecture? Thats insane.

NV50 ready in three years for the XB2 launch I see as more likely then not.

Well for nVidia's sake, lets home your hardware predictions are better than your software ones

 

[megadrive0088]

The XGPU is essentially an NV25 (more or less). The PC version (GeForce 4 Ti) launched aproximately Feb 2002, or about 4 months after the xbox.

XGPU is a bit less than the NV25, but more than the NV20. I think of it as a "NV23"

So given NVidia's patterns (assuming NVidia maintains their 6 month cycle, and NV30 launches in January), we can probably expect them to be at something resembling an NV50 or NV55 by the time we hit the end of 2005, which is when any console launch for 2005 would be planned.

Presuming xbox2 follows the pattern set by xbox1, initial studies will occur in the second half of 2003, design work will start in early 2004 with lockdown in early 2005, and the final product will release in late fall 2005 with something resembling an NV55.

sounds about right to me. I mean, none of us really knows what Nvidiia's scheduel will be like over the next 2-3 years, but your estimate is something I would go along with. I buy it

Note this is all highly speculative (ie. pulled out of my ass ) and relies on many things which could change at any time, like NVidia's release schedule and chip numbering schema, or ATI being chosen as the supplier, or even Microsoft deciding to go with a custom architecture and designing their own chips.

All very true. but at least you are taking into account most of the major possibilities. I think MS going with ATI is a much greater possibility than MS designing their own chips, but there is a slight chance of that happening. my best bet is still on Nvidia though, with all the patents, IPs and engineers (3Dfx, GigaPixel, some SGI) they have... Although ATI has a hell of alot now too (ArtX, some GE/Real3D) It'll come down to who can deliver the most impressive tech for the fewest dollars. ATI has a real shot IMO. Though Nvidia is still probably the frontrunner.

 

[PiNkY]

Remember that the NV2A was several months ahead of its PC counterpart. If the XB2 were to launch today it is likely it would ship with NV30(MS would find a fab that could pull it off costs be damned, I think we have seen that enough). If need be, I could see them approaching Intel or AMD and using their fab capabilities(Dresden has plenty of room ATM) tied in to landing the CPU for the XB2.

I really do not think that was the case. For once NV25 could arguably have been released two - three months earlier (there was an article on EETimes regarding its dev. cycle). IMO the reason they did not release it last years christmas were their GF3 TI line of gpus, which became their fall part as a response to ATIs R200. There were interviews in Summer 2001, where Nvidia employees stated that their fall part would be an architecturaly new thing (i.e. nv25), yet with the release of ATIs R200 they probably favoured going with a faster clocked Gf3 (i.e. TI) and keeping the performance crown, instead of leaving that perk to ATI for the two months it would have taken to get NV25 out of the door. And nv2a was released well after GF3, which is certainly the tech. generation it and nv25 belongs to as changes were rather minor (about a five percent increase in trans. count, addition of a second vertex shader, and some reorganizations of their piplines TMUs). BTW how does XBOX perform in regards to aniso, more in the way of GF3(rel. small hit), or like GF4?

Regarding graphical capabilities for ng consoles, i suppose that the diffrences will be rather technical in nature, for all 2005-2006 architectures will probably have the ability to render at PAL/NTSC resolutions with all bells and whistles (10+ passes per pixel, AA, and AS). What i mean is that at that time to have visibly superior graphics for John Public (@ standard res.) one of the vendors would have to have a serious advantage in graphics processing power (2x - 3x) over the competition which is probably not going to happen (E.g. take radeon9700 to gf4-TI4600, while all us geeks would probably agree that that the ati card is lots more powerfull for dx8 content, showing UT2003 running at 1280*1024on gf and 1600*1200 (both @4x AA) on rad. would not make John P. think that this a difference like night and day, but rather that a latter looks subtly improved.)

 

[bukima]

SONY is sitting on their pearls, idling around, while MS is quickly adopting fast developing PC technology for the console market. MS can cheaply benefit from huge R&D investments by PC semiconductor companies like AMD, Intel, NVIDIA and ATI, while SONY thinks they're able to do all themself with a little help by IBM, a software/service company.

Don't make the mistake of thinking the console market can be made to work like the PC market. Yes, IE won over the world with frequent "sequels", but it was and is free. No one is going to complain about downloading a new browser every so often when it doesn't cost anything...

Consoles on the other hand, are a very expensive part of the console gamers budget. Only a tiny hardcore group will be willing to buy a console if it will only last you three years.

Besides, we have already seen that specs or launch dates don't give you any significant sales. Brand value is what sells your console these days - and Sony has tons of it, Xbox has next to none. And releasing Xbox2 early will only chip away at what little there is, especially outside the US where few give a green damn anyway.

Sony is sitting very pretty for the next gen also, and MS can only hope to keep a foot in the door and slowly build up their brand. To do this, they need to offer value in all departments - price, games, shelf-life etc.

 

[megadrive0088]

Even if NV25 had been released as PC cards by fall 2001, the NV2A was still a little bit ahead of it. NV2A taped out in Feb while NV25 was at least a couple months later, spring , summer or fall tape out, don't know exactly when though.



As for XBox2, since NV30 is done right now, Nvidia has engineers working on NV35 and NV40 at this moment. I'm sure they've been working on NV40 for at least a few months now. I am sure that sometime in 2003, perhaps when another development team is freed up (not the NV40 one)
that NV50 will enter serious development. While everything depends on how fast Nvidia is developing new chips, it's almost certain that an NV4X or NV5X will form the basis of XBox2 graphics.

Perhaps XBox2 will use a radically enhanced NV4X or NV5X, or perhaps multiple GPUs. Since PS3 is going to be obsenely powerful with many, many CELLs in parallel, and no doubt many pipelines as well, I would favor MS/Nvidia using 4 of the most advanced GPUs they can come up with in XBox2. Perhaps 4 enhanced NV45s or enhanced NV55s. I doubt Nvidia is going to have a single chip by 2005 that can outperform the PS3. While I believe Nvidia (and ATI) will be significantly ahead of Sony in terms of image quality and feature set, Sony will likely have a large advantage in bandwidth and parallism with EE3+GS3 in PS3.

If Sony is going to have massive parallelism in PS3, why not have at least a high degree of parallelism in XBox2. ATI's R300 was design with massive parallel processing in mind. you can combined upto 256 R300 GPUs in one system. NV30 probably has a similar capability. Even though Nvidia has never favored multi chip approach (and I still think NV30 is one chip btw) times are changing.

IMHO todays concoles are much too small, and have far too little memory.
even XBox is pretty small compared to where I think they should be. consoles should be the size of VCRs or DVD players. Think about how much more room you'd have to put in memory and other processors, WHILE STILL being at consumer price points. Just because you have a larger piece of equipment like a DVD player doesnt mean its expensive. DVD players go for at little as $99, perhaps less now. I have a feeling that PS3 and XBox2 will be larger than their predasessors. Tehy're both likely to have 1GB or more memory. With PS3, I dont see all of that fitting
into a box the size of PS2 with all of those CELL chips inside as well.

Since both PS3 and XBox2 are aiming to become the hub of the living room (I know, whasn't this suppose to happen with Ps2 and Xbox?) I see the consoles growing in size conciderabley. There should be enough room in XBox2 to have 4 GPUs a CPU, audio chip, HDD, and lots of memory. An elegant example of multi chip processing would be the NAOMI2 with its twin PowerVR2DCs rasterizers + ELAN T&L + SH-4 CPU + over 96 MB memory. Even 3Dfx had single PC cards with 4-8 or more chips on them. XBox2 will be conciderably larger than a PC card, even XBox1 is. And MS can easily easily swallow the cost of multi GPU XBox2, they will almost have to if they want to stay in the race with Sony.

 

[BoddoZerg]

Applying the PC model to console games *might* work if MS keeps the DirectX framework for complete backward and forward compatitibility.

That is, if three years from now there is an Xbox, an Xbox2, and an Xbox3, and every game that runs on one of the Xboxes can run on all the other xboxes (at highly different graphics settings - just like any PC game) then it might actually be a viable business model. People wouldn't have to upgrade directly from an Xbox bought in Christmas 2001 to an Xbox2 in Christmas 2003 and an Xbox3 for Christmas 2005, because any game that runs on xbox2 would be exactly the same on Xbox, just lower graphics settings.

Of course, this is pure speculation, based on my admittedly PC-slanted worldview.

 

[Megadrive1988]

An XBox3 is many, many years away. I cannot see MS bringing out a new XBox every 2-3 years. maybe every 4-5 years. consoles need a life cycle.
2 or 3 years isn't really a life cycle because the best games usually come in years 3 and 4 of a console's life.

XBox2 might be out by late 2005. XBox3 maybe by 2009-2010

 

[archie4oz]

IMHO todays concoles are much too small, and have far too little memory.

People always say this, yet don't realize no matter what happens next, they'll continue to say systems don't have enough memory. Coming from systems with only a couple of MBs of RAM to today's systems feels like taking in a fresh breath of air. Today's sytems have 8-20 times the memory of the previous generation, do you honestly think that just adding more memory will make things better? It'll just get used up and you'll be saying the same thing about the following systems needing more memory...

even XBox is pretty small compared to where I think they should be. consoles should be the size of VCRs or DVD players.

Really? You do realize that not everybody has big houses, or lives by 19" rack standards. Some also like to pack their console in their backpack to take to their friends' place, or grandma's. BTW, some DVD players can be quite large (weighing in at 10-15kg) and can be quite small. You statement does seem rather broad...

Tehy're both likely to have 1GB or more memory. With PS3, I dont see all of that fitting
into a box the size of PS2 with all of those CELL chips inside as well.

You do realize that there are DIMMs of 2GB capacity today right? Memory of all things is the first to benefit from process advances. Regarding chip size, even a 180nm I-32 GS fit rather easily in the palm of my hand, and it contained more than twice as many transistors as NV30 is predicted to have, and EE+GSs are coming out in single chips nowadays. There's plenty of room to go still...

While I believe Nvidia (and ATI) will be significantly ahead of Sony in terms of image quality and feature set,

I'm not sure image quality is going to be something that's going to be a distinguishable factor in a couple of years. It's rather easy to generate very high image quality with today's hardware. We're getting to the point of diminishing returns, and the content (or how much of it can we cram) of what we display is going to be of more critical importance than how we display it. Likewise audio may finally get it's time in the limelight (as audio has become more so lately. And finally AI (which has become probably the most critical and busy topic as of late), and how it affects gameplay will become the most critical factor...