NVIDIA Roadmap Outline for 1H08

I wouldnt count on seeing a 9800gtx. Silus seems to live in a world of his own when it comes this imaginary card and is just making up his own specs. there is no evidence that there will be a 9800gtx.

First of all, that's right. I am giving my opinion on what the 9800 GTX will be. I'm betting on a 192 SPs card, but it having 160 SPs, seems to be a very good possibility. And it will most likely use the same 384 bit memory interface in G80.

Second, you can quote me on two things:

1) There WILL be a 9800 GTX
2) It WON'T just be an overclocked 8800 GTS 512

As for G92, you and others, keep spinning what I said. I never said the 9800 GTX was going to use G92. You and others did. It will be based on G80/G92, but it will probably not have the same codename.
 
NVIDIA_G92.DEV_0612.1 = "NVIDIA G92-420"

This text can be found on Nvidia drivers.. wonder what this card will be?

8800 GTS 512/1024 = G92-400
9800 GX2 = G92-450

So it could be 9800 GTX.
 
NVIDIA_G92.DEV_0612.1 = "NVIDIA G92-420"

This text can be found on Nvidia drivers.. wonder what this card will be?

8800 GTS 512/1024 = G92-400
9800 GX2 = G92-450

So it could be 9800 GTX.

Or it could be a 9800 GT. ;) (Which is basically just a tri-SLI capable 8800 GT if you believe the article.)
 
I'll doubt that. It would mean that Nvidia would change their code-name strategy: Bigger number in the end; better performing card

G92-270 8800 GT
G92-400 8800 GTS
G92-450 9800 GX2

Now G92-420 9800 GT which would lose against 8800 GTS?
 
At the top-end of the 9-series cards will sit the dual-GPU GeForce 9800 GX2 which will launch on March 11 this year. This will be followed by the GeForce 9800 GTX and 9800 GT between the end of March and beginning of April.
Source

This pretty much confirms the HardOCP article.

According to these sources (1, 2, 3) the 9800 GTX will only be a higher clocked 8800 GTS 512.

Other cards will follow: 9600 GS, 9500 GT, 9500 GS.

I wonder why Nvidia is establishing this new series so close before the launch of their GT200 card (June/July).
 
Source

This pretty much confirms the HardOCP article.

According to these sources (1, 2, 3) the 9800 GTX will only be a higher clocked 8800 GTS 512.

Other cards will follow: 9600 GS, 9500 GT, 9500 GS.

I wonder why Nvidia is establishing this new series so close before the launch of their GT200 card (June/July).


I dont think there is anything in there blazingly wrong. I should have a update early next week as well.
 
I don't think the G200 will be released this year unless the 770 by ati outperforms the 9800GTX... if that does not happen we will not see the G200 for another year.
 
ATIs new highend card will be a RV770 X2 and this won will be significantly faster than the 3850 X2. So Nvidia has to release mor than a Dual-G92 and a higher clocked GTS (GTX).

Preliminary reports from nVidia and preliminary reports about RV770X2 indicate that the 9800GX2 may very well already be at the performance level of RV770X2 (add in some mature drivers, lower prices, and OC versions by the time RV770X2 arrives and nVidia may already have its RV770X2 competitor). However, I have also seen reports from manufacturers like eVGA that indicate nVidia plans a new project launch in June/July. Preliminary rumors about those indicate that they will be of only incremental increase over the 9 series (they may well be the 9850's or 9900's)- or they may just be a die-shrink of the entire 9 series rather than a new core. Anyway, it does not seem like nVidia will be launching its true "revolutionary core" that has been in R+D for quite some time now until at the very earliest this year's end. Why not? ATi doesn't seem to be giving them much of a reason to launch it at all (my guess is the June/July cores will just be 55nm versions with larger memory buses and more memory).
 
Do you have any idea why this series is launched so close before the R700 & GT200 release?


ATIs new highend card will be a RV770 X2 and this won will be significantly faster than the 3850 X2. So Nvidia has to release mor than a Dual-G92 and a higher clocked GTS (GTX).

Well it will be a while before we se 770x2's I mean correct me if I'm wrong but they have yet to release the 770 and are still have significant issues with the 3870x2 driver, so before the 770x2's they need to hammer out the last version. Or again am I missing something?
 
Preliminary reports from nVidia and preliminary reports about RV770X2 indicate that the 9800GX2 may very well already be at the performance level of RV770X2 (add in some mature drivers, lower prices, and OC versions by the time RV770X2 arrives and nVidia may already have its RV770X2 competitor). However, I have also seen reports from manufacturers like eVGA that indicate nVidia plans a new project launch in June/July. Preliminary rumors about those indicate that they will be of only incremental increase over the 9 series (they may well be the 9850's or 9900's)- or they may just be a die-shrink of the entire 9 series rather than a new core. Anyway, it does not seem like nVidia will be launching its true "revolutionary core" that has been in R+D for quite some time now until at the very earliest this year's end. Why not? ATi doesn't seem to be giving them much of a reason to launch it at all (my guess is the June/July cores will just be 55nm versions with larger memory buses and more memory).

QFT, this mirrors exactly what I was saying! Just probably a lot better :)
 
Preliminary reports from nVidia and preliminary reports about RV770X2 indicate that the 9800GX2 may very well already be at the performance level of RV770X2 (add in some mature drivers, lower prices, and OC versions by the time RV770X2 arrives and nVidia may already have its RV770X2 competitor). However, I have also seen reports from manufacturers like eVGA that indicate nVidia plans a new project launch in June/July. Preliminary rumors about those indicate that they will be of only incremental increase over the 9 series (they may well be the 9850's or 9900's)- or they may just be a die-shrink of the entire 9 series rather than a new core. Anyway, it does not seem like nVidia will be launching its true "revolutionary core" that has been in R+D for quite some time now until at the very earliest this year's end. Why not? ATi doesn't seem to be giving them much of a reason to launch it at all (my guess is the June/July cores will just be 55nm versions with larger memory buses and more memory).
I don't think this scenario is very likely. The R700 (ATIs Dual RV770) is said to be on schedule for June/July. In addition it's supposed to be at least 50% faster than the X2.

The 9800 GX2 will be like 8800 GTS SLI with a higher clocked memory.

At the moment I believe the GT200 will not be a total new series. Rather something like a G90: A G92 with a bigger memory interface, more SPs and so on.
 
Hmm.. Since the 9600GT launch, we are seeing that new Compression Technology deliver a solid 15% increase in a clock-for-clock situation. So with that in mind.. a Small bump on a g92 Core in speed and this compression could net a nice increase in speed without increasing power consumption or heat...

So the "9" series does offer something "new" that is of great value..
 
Hmm.. Since the 9600GT launch, we are seeing that new Compression Technology deliver a solid 15% increase in a clock-for-clock situation. So with that in mind.. a Small bump on a g92 Core in speed and this compression could net a nice increase in speed without increasing power consumption or heat...

So the "9" series does offer something "new" that is of great value..

Yeh, I was quite impressed by the reviews laid down on the 9600GT. They're getting close to 8800GT speeds on a seemingly much lesser card.
 
How can y'all say that th 9 series is gonna suck? The budget 9600GT was pretty good from what I saw, and yet on paper it looked like it would compare to the 8600GTS. That new compression tech they got and whatever else new they got is gonna help the 9 series. I think the 9800GTX is gonna rock. It's gonna be at least 20% faster than the ultra.
 
Why not? ATi doesn't seem to be giving them much of a reason to launch it at all .

Oh gimme a break with this junk. Nvidia wants to STAB AND KILL AMD anyway they can. That's business. And you're telling me they're somehow holding off until AMD launches something? Once again, GIVE ME A BREAK.

You think Intel is going to hold off Penryn and all that crap? Why not, it's the same situation. Intel already has the clear performance lead, why dont they just hold off until AMD comes up with something better then? The absurdity of imaginig Intel doing that ought to illustrate the absurdity of thinking Nvidia is.

Lets illustrate with another example, Say Nvidias next architecture is ready, but AMD's is six months away (actually that sounds familiar, Nvidia already launched a new architecture six months before AMD TWICE in the recent past, yet more proof your theory is ridiculous). Say it's twice as fast. Why on earth WOULDNT they launch it now, what could it possibly harm? It would only harm AMD because now Nvidia has something even faster and nobody would buy an AMD card at all. Hell, Nvidia launching way sooner was a lot of what killed ATI the last couple gens! I mean right now Nvidia is selling bigger and more expensive (8800GT) cards than AMD for $200, not leaving a lot of profit. Dont you think they wish they had something twice as fast to sell for $500? Dont you think that would only help their profits and also hurt AMD?

Another thing, AMD already launched it's dual GPU card, why isn't the Nvidia counterpart out? We know it's coming very soon, so they're obviously not holding it back..

And another thing, right now in certain cases, the fastest single card on the planet is 3870X2. In Quad crossfire also, in certain cases Nvidia has NOTHING as fast since Nvidia can only max at two cards SLI currently. So hell, by your logic, maybe AMD is the one holding back, waiting for some Nvidia competition?

Nope, the fact is Nvidia has been suffering delays lately.

Nvidia doesn't have anything, dont make excuses. Amd doesn't really have anything either, but obviously that's not the point of this rant.
 
How can y'all say that th 9 series is gonna suck? The budget 9600GT was pretty good from what I saw, and yet on paper it looked like it would compare to the 8600GTS. That new compression tech they got and whatever else new they got is gonna help the 9 series. I think the 9800GTX is gonna rock. It's gonna be at least 20% faster than the ultra.

Hmm Keep my 8800GTX at Ultra speeds = Free
Pay $500-600 for 20% fps increase... so in Crysis 30fps goes to 36fps wow... 6fps that is approximately $100 per fps...

Um no thanks... Now when the G200 comes and you compare that with 60fps in crysis (estimate obviously), but 30fps for 600bucks is $20 per second which is a bit more tolerable.
 
You think Intel is going to hold off Penryn and all that crap? Why not, it's the same situation. Intel already has the clear performance lead, why dont they just hold off until AMD comes up with something better then? The absurdity of imaginig Intel doing that ought to illustrate the absurdity of thinking Nvidia is.

Good example there, and Intel is holding off Penryn a bit thanks to AMD. Namely, they're holding off the quad-core Yorkfields (Q9xxx). They want to clear out their stock of Kentsfields first, and thanks to Phenom not being much of a threat, they have the leisure to do that. They didn't hold back the Wolfdales though, obviously.

Analogies to NVIDIA: It's doubtful that NVIDIA would hold back G100 aka GT200 completely just due to no competition from AMD. Doubly so because they do have competition from AMD now, much more so than Intel does processor-wise. The 3850/3870 are selling like hotcakes. The only think AMD doesn't really have competition for now, graphics-wise, is the single-GPU ultra-high-end, and that's more for bragging rights and the halo effect than where most of their profits are made. That said, since NVIDIA is still unquestionably in the lead, like Intel they have the privilege to wait just long enough to clear out old stock before releasing the new chips, just like Intel. It would not be good business practice to hold out until AMD catches up though, like you said.
 
Wow some people need to head back to business school.. its called extended a product life cycle. It means they get more $$$ for a sunk cost investment. Almost all companies do it, especially if they already have a competitive advantage.
 
Thank you Sunin for stating the glaringly obvious.

It costs money to switch over to a new fab process.. and every change they want to milk as much profit as possible from that while at the same time meeting or exceeding market expectations. Everyone in the know knows right now Nvidia still has the crown. They have grown comfortable with the 8800 series and yet have made some changes to increase yield and deliver a better product. As production gets cheaper and prices stabalize profit on a per unit basis just keeps growing.

That is why...

it is also the reason it took so long for cdrom drives to hit the 50x speed ratings we have today that are common. They let each speed have some time in market before releasing the next to keep profit up.
 
Right- nVidia is already exploring 45nm and 55nm manufacturing processes, but G92 and G94 are 65nm manufacturing processes. Undoubtedly, their GT200 core will be 55nm or 65nm, thus forcing them to switch manufacturing processes. That's an expensive thing to do. So, if they just continue to improve their current cores... And, I should also mention that I'd imagine die-shrinking from 65nm to 55nm or 45nm would present them with fewer technical issues because there should be fewer variables. Therefore, if they can die-shrink an existing architecture, it should theoretically allow them to adapt to that smaller process a bit better in order to prepare for a new architecture. Given that from rumors I've heard, the stop at 65nm seems like it will be a relatively short one for nVidia.

Anyway, the point is that nVidia doesn't have to release a completely new architecture and go through the expense that entails. If they can put in just a relatively small amount of R+D to improve their current architecture (and, I'm sure some of that R+D will also aid them w/their newer architecture) they can continue to pour R+D into the new architecture, continually improving it. Once they find they can't take their current architecture any further, then I feel you will see GT200 or whatever the hell it is.

As I recall, the GeForce 8 series was developed over a ~5 year span, beginning in the GeForce 5 era? Anyway, fact of the matter is that they came out with the GeForce 6 series and gave ATi some good competition. They improved on the GeForce 6 architecture w/the GeForce 7 series and continued to compete w/ATi. At this point, nVidia realizes that it's time to bring out the unified architecture (ostensibly they were getting a little desperate to compete w/ATi by the time of the 7950GX2- I think part of the reason for the present 9800GX2 is that nVidia feels they put together a desperate pile of crap w/the original and want to revisit the GX2 concept from a different viewpoint) as they're at the end of the line for their current architecture and ATi's x1950 is giving them some problems. So they come out with the 8 series- when they needed it.
 
All I know is that I've been using this 8800gts 640 for going on 16 months now and still not a new generation in sight. I don't think in the 11 or 12 so years of 3d acceleration that I can remember a stretch of time that long where I've been using the same video card. Something is definitely wrong in the market if they are releasing a new generation of video card at 2 year spans. Games still run great on this card but it is odd that there is no sign of a new generation of card anywhere on the horizon.
 
maybe the fact that pc gaming is getting less and less popular due to these ps3's and xbox 360's taking over could be an issue too. there are some major stores like HMV who literely stopped selling pc games period

if games are being focused more on consoles then no wonder why pc's are having an effect. no wonder why your 16month old card can still play almost every game without breaking a sweat.

besided crysis what other games merit a reason for nvidia and ATI to work like a dog producing another ground breaking gpu card? these are truely sad times for pc gamers
 
maybe the fact that pc gaming is getting less and less popular due to these ps3's and xbox 360's taking over could be an issue too. there are some major stores like HMV who literely stopped selling pc games period,

Not really if you think about it what do the Xbox 360's and PS3's use to do the graphics... one has nvidia one has an ati... its still in their best interest to be top dog.
 
Not really if you think about it what do the Xbox 360's and PS3's use to do the graphics... one has nvidia one has an ati... its still in their best interest to be top dog.

yea top dog for the console format but not for pc format. i mean why do they make gpu cards? for games right? And like i said before, if games are mainly focused on consoles then there is no need and no rush to work there socks off making a true next gen gpu card for us pc users because most of the games are for consoles and we all know that consoles are far weaker then pc's technicaly so when you see a port of say umm rainbow six vegas 2 or even gta4 to the pc, even a 7900 card could probably run those games at decent settings and how old is that card?

what we need is these game devs to make proper pc games that fully utilise 8800 series and behond like what crysis has tried to do and then gvie nvidia/ATI a reason to push and release a proper next gen card. has it stands, most pc games that will come out some time year will be able to run very high on your 16 month old card
 
what we need is these game devs to make proper pc games that fully utilise 8800 series and behond like what crysis has tried to do and then gvie nvidia/ATI a reason to push and release a proper next gen card. has it stands, most pc games that will come out some time year will be able to run very high on your 16 month old card

To sell to the very small percentage of gamers who have 8800's? It's a vicious cycle. Most people don't have good graphics cards (or worse, are stuck on integrated because they don't know any better), so game developers won't push the envelope. Game developers don't push the envelope, so the graphics card companies don't have much motivation to get their newest tech out there faster. What we need is Intel/NVIDIA/AMD to make integrated graphics worth something, so that anybody with a computer can play even the latest games (albeit on lower settings), and maybe then PC gaming will begin to overtake consoles.
 
As far as stores not selling PC games anymore, that's thanks to digital distribution- not to any kind of declining PC popularity. Simply, if I can pay the same amount for a game, save a trip to the store, run that game without needing to have the CD/DVD in my drive, and be able to relatively quickly (Sins of a Solar Empire was dled and installed inside of 10 minutes) download that game to the PC I am presently using without needing to have the CD/DVD present, then I'll certainly opt for that rather than the physical copy. The only benefit of a physical copy is potentially price because presently there is more competition among physical copy retailers than there is among digital distribution retailers and having a physical copy of the manual (which is becoming less and less important).

Anyway, the 360 is shooting itself in the foot thanks to its 30% failure rate- I have friends who previously scoffed at PC gaming who are giving it a serious look-over thanks to multiple 360 break-downs. Plus, PC gaming (if you buy your hardware in moderation) generally yields a superior experience for ultimately less money (you can get an off-the-shelf PC from Gateway right now that costs $1100 and will run the first half [the only half anyone cares about] of Crysis on High settings; if you bought a PC you weren't going to use for gaming, it would cost your between $600-700 to get something decent, so that's only a ~$400 difference, which cheaper games and the like will easily make-up for; either cheaper games or mods, b/c mods will extend the lifetime of your games thereby satiating some of your need for new games therefore causing you to ultimately buy fewer games). But I must say- cards like the 9600GT and 8800GT are going a long way towards making PC gaming a much more accessible option for many.
 
As far as stores not selling PC games anymore, that's thanks to digital distribution

That has nothing to do with digital distribution. You are confusing cause and effect. Digital distribution is an industry response to both getting more control over distribution for the developer and also waning PC sales through normal distribution.

Only a very small portion of PC games have good digital distribution and even fewer are successes aside from what Steam has done. PC game sales in brick and mortar stores are down because of low demand, piracy, the inability of your average PC to play the groundbreaking titles well, and simply a lack of interest due to the feedback loop of lack of good titles and variety compared to consoles.
 
That has nothing to do with digital distribution. You are confusing cause and effect. Digital distribution is an industry response to both getting more control over distribution for the developer and also waning PC sales through normal distribution.

Only a very small portion of PC games have good digital distribution and even fewer are successes aside from what Steam has done. PC game sales in brick and mortar stores are down because of low demand, piracy, the inability of your average PC to play the groundbreaking titles well, and simply a lack of interest due to the feedback loop of lack of good titles and variety compared to consoles.

The NPD itself has stated that digital distribution is a major cause of declining retail sales- link

Well, I suppose I should probably provide some additional quotes and links to prove what I'm talking about here with the PC still being a very financially strong platform:Of course, the main reason why PC gaming is perceived to be doing horribly compared to the consoles is because of the difference in how PC games sell compared to console games. First, PC games generally don't make most of their sales in their first few weeks. Console games do. Why don't PC games do this? They don't get as much advertising (so they also have a lower overhead meaning they can turn a profit with fewer sales than console games generally can)- even Crysis didn't get a great deal of advertising- PC gamers seem a little more laid-back as far as purchasing patterns go (partly because a lot of your really good PC games will last you a lot longer thanks to mods and quality online services), and many PC gamers will wait to buy games until their next upgrade (so, yes, that is a factor, but only a factor in the immediacy of sales). In the long run, PC games tend to sell just as well, if not better, than console games. Second, the PC is more globally prominent than consoles are, and thus PC games make more international (read- non-US) sales than console games generally. This factor also contributes to why PC games sell more slowly. And, finally, the NPD is generally used as the most prominent indicator of how well a game/platform is doing. Given that NPD analysts themselves acknowledge its present flaws in relation to gaging PC sales, it's no surprise that it really isn't a great indicator for PC games. Here are a few reasons why:
  • The NPD does not cover digital distribution/many online sales
  • Video cards (huh? Yeah- what game comes with many new video cards nowadays? Crysis. What game would most people wait to buy until they upgraded their PC's? Crysis. Therefore, you have a lot of people getting Crysis included in their video card purchases, and thus not buying it from a source the NPD can reflect in its sales figures; this is how I got Enemy Territory: Quake Wars, which is a game I might have purchased had it not come w/my 8800GTS SSC; had I not already purchased Crysis, I would have gotten the SSC package that came w/Crysis, b/c I actually had an option)
  • The NPD only reflects sales in one of the most console-dominant regions on the globe
  • And anything I missed, this article covers as it delves into the misconceptions about Crysis and UT3 sales
And, finally, if digital distribution and PC gaming are not doing well, why would Amazon (who already sells PC games and definitely has a good idea of how much revenue they constitute) be expanding into digital distribution?

The fact of the matter is that digital distribution is rising in prominence and is cutting into retail sales, and that is the major reason you see a 6% decrease in PC gaming's NPD.
 
IMO digital distribution took off when those going to a store noticed limited choices.
The limited choices started years ago. Hard to purchase retail copies when they aren't carried...
 
IMO digital distribution took off when those going to a store noticed limited choices.
The limited choices started years ago. Hard to purchase retail copies when they aren't carried...

(shrug) I've yet to buy a published game digitally because I could not find a physical copy (except for Act of War: High Treason because that was released in Europe first and Direct2Drive got the Americanized version 2-3 months in advance of retailers). Most of the time, I'll make my digital purchase and walk into the local Target (yeah, Target- not Gamestop or any kind of gaming-orientated store- TARGET) and see the physical version sitting on the GFW endcap.
 
Without seeing internal company documentation of the major publishing, development and hardware studios, speculation remains speculation.

However, for those raging against Nvidia's recent failure to innovate, consider that, aside from the obvious sunk cost argument, there is no title out there, aside from one, that even taxes the 16 month old G80 architecture.

Now we don't know whether that's because Nvidia isn't innovating in the first place, but we do know this: the money to be made in gaming lies with consoles, and the major developers like to develop concurrently for the key platforms with the big titles (360, PS3, PC), or at least be able to port easily. They have no interest in making life more difficult for themselves by adding fancy sharply different features for the relatively small PC market.
 
well said MAT. all developers will do is simply port console games into pc games hench why the majority of pc games that will come out this year, your 16month old gpu card will still be able to run it. now if i was nvidia why would i want to rush and make a true next gen card when our old tech is still running most games without breaking a sweat?
 
Without seeing internal company documentation of the major publishing, development and hardware studios, speculation remains speculation.

However, for those raging against Nvidia's recent failure to innovate, consider that, aside from the obvious sunk cost argument, there is no title out there, aside from one, that even taxes the 16 month old G80 architecture.

Now we don't know whether that's because Nvidia isn't innovating in the first place, but we do know this: the money to be made in gaming lies with consoles, and the major developers like to develop concurrently for the key platforms with the big titles (360, PS3, PC), or at least be able to port easily. They have no interest in making life more difficult for themselves by adding fancy sharply different features for the relatively small PC market.
well actually there is at least two since COJ in DX10 is a system killer.
 
Hmm what are the #'s of consoles? I'm Highly suspicious of the arguement that the PC market is small...

Considering a game like WOW has 10million subscribers alone it makes you wonder how many PCs are truly built for gaming...

Or are you saying small pc market meaning the Enthusiast that gets the best cards upon release? Please clarify a bit.
 
Back
Top