Fudzilla: ATI/Nvidia DX11 cards will come in 2009

[Kakkoii]

Nvidia:

Fudzilla Tuesday, 23 June 2009 13:44

Our sources have confirmed that Nvidia is working hard to launch its first DirectX 11 in this year. Well informed sources are telling us that the plan is to launch in late Q4 2009, but they can still make it before the end of the year.

Many people were reporting that Nvidia's high end GPU, something that we all call GT300, might be delayed for early 2010. Well we can tell you that this won?t happen, but we strongly believe that Nvidia cannot make it in time for the Windows 7 launch.

ATI has every chance to launch its DirectX 11 before Nvidia. Nvidia will come just a month or two later, but still in 2009.

ATI:

Fudzilla Tuesday, 23 June 2009 13:49

We've learned that ATI's upcoming mainstream DirectX 11 part, something that we call RV870, should be ready for the Windows 7 launch.

ATI might even launch its card earlier, but if they align the launch with Windows 7, ATI should be launching its DirectX 11 performance card in the second half of October.

This is at least timeframe that ATI plans for its hardware launch and we are quite sure that ATI should launch DirectX 11 before Nvidia. At the same time, we?ve heard that ?RV870? is not really a brand new architecture, and that you should expect more shaders, DirectX 11 support but all based on RV770 concepts.

Naturally, the new ATI chip is 40nm which enables higher clocks and more transistors and shaders but so far we don?t have any numbers we can quote.

 

[Scali]

Hum, it would be a shame if the RV870 is indeed just an RV770 pimped up to DX11 spec. That seems to imply that there will not be a considerable performance leap in general, not much beyond what you'd expect from a regular die-shrink anyway.

 

[Wreckage]

Originally posted by: Scali
Hum, it would be a shame if the RV870 is indeed just an RV770 pimped up to DX11 spec. That seems to imply that there will not be a considerable performance leap in general, not much beyond what you'd expect from a regular die-shrink anyway.

They will get destroyed in the next round if the rumors are true. They already took a bad enough beating this round, I'm not sure if they could afford such a catastrophe.

I was hoping for something new and another round of competition.

 

[OCGuy]

Yea if 5XXX it is just a refresh, my GT300 is going to be very expensive.

Of course, everything here is just rumor. ATi has been keeping quiet.

 

[TheMeanestGuest]

As has been proven in the past, you can't really know how a card will perform until it's actually launched (or very shortly before!).

It's entirely possible both camps will launch products that are somewhat lacking, or they could both be amazing or one could clearly be superior to the other. So yeah, pretty much anything ;Q.

I haven't ever really payed too too much attention to pre-launch rumours before, does anyone happen to recall any that were actually spot on?

 

[Lonyo]

I don't see the HD5xxx being that much more than a slightly tweaked HD4xxx with more shaders and DX11 spec, but that's not to say there won't be a performance leap.
More shaders + more clock speed = more performance. Not competing at the high end is something ATI have in their game plan, so I wouldn't be surprised to see once more affordable, adequately performing cards.

 

[KingstonU]

Originally posted by: Wreckage

Originally posted by: Scali
Hum, it would be a shame if the RV870 is indeed just an RV770 pimped up to DX11 spec. That seems to imply that there will not be a considerable performance leap in general, not much beyond what you'd expect from a regular die-shrink anyway.

They will get destroyed in the next round if the rumors are true. They already took a bad enough beating this round, I'm not sure if they could afford such a catastrophe.

I was hoping for something new and another round of competition.

I am reading this correctly?? You're saying ATI took a beating this round???? Nvidia has been the one getting completely destroyed in every segment except for the absolute highest end this entire 4xxx vs GT200 round since introduction. I can't recall a single buyer's recommendation for an Nvidia card in any segment except the very high end. (260 core 216 was close to 4870, and 275 was close to 4890, however those equalities were short lived as ATI has the advantage of being able to drop prices faster and get the advantage)

 

[Barfo]

Originally posted by: Wreckage

Originally posted by: Scali
Hum, it would be a shame if the RV870 is indeed just an RV770 pimped up to DX11 spec. That seems to imply that there will not be a considerable performance leap in general, not much beyond what you'd expect from a regular die-shrink anyway.

They will get destroyed in the next round if the rumors are true. They already took a bad enough beating this round, I'm not sure if they could afford such a catastrophe.

I was hoping for something new and another round of competition.

And another thread goes down the drain, well done fanboy!

 

[Fox5]

If ATI launches a few months early with something that beats nvidia's current top-end, that might just be game and set. The high end ATI cards will scoop up all the early upgraders, and nvidia's gigantic dies won't be able to drop in price enough to pick up the mainstream.
It's what nvidia did with the geforce 6 and 7 series, and even though ATI had the overall faster card (way faster in the case of the 7 series against the x1950), nvidia won those generations easily.

 

[OCGuy]

Originally posted by: KingstonU


I am reading this correctly?? You're saying ATI took a beating this round???? Nvidia has been the one getting completely destroyed in every segment except for the absolute highest end this entire 4xxx vs GT200 round since introduction. I can't recall a single buyer's recommendation for an Nvidia card in any segment except the very high end. (260 core 216 was close to 4870, and 275 was close to 4890, however those equalities were short lived as ATI has the advantage of being able to drop prices faster and get the advantage)

Your personal experience in seeing recommendations does not mean that is what is happening in the market place.

 

[brblx]

both companies have 'taken a beating.' neither has done well recently- the current market is good for the consumer, but both companies have had to take a hit in profits to keep competitive.

but the fact of the matter is that nvidia was forced to practically halve their GPU prices in a 6 month period because of the release of the 4xxx cards. i would say that is significant. and ati will probably do the same thing with their next gen cards by continuing to pursue the value market more than the uber-$500-card crown.

 

[SlowSpyder]

Originally posted by: Wreckage

Originally posted by: Scali
Hum, it would be a shame if the RV870 is indeed just an RV770 pimped up to DX11 spec. That seems to imply that there will not be a considerable performance leap in general, not much beyond what you'd expect from a regular die-shrink anyway.

They will get destroyed in the next round if the rumors are true. They already took a bad enough beating this round, I'm not sure if they could afford such a catastrophe.

I was hoping for something new and another round of competition.

You didn't seem to mind when Nvidia did that with their current cards... just the same, but more of it. With both companies building on smaller tech they could both add more 'stuff' and possibly add clock speed... while hardly revolutionary both could certainly offer a jump in performance over what we can get today. But then again how much faith do you want to put into rumors? Rumor was the 4870 was supposed to only compete with the 9800GTX, but we all know how that turned out.

 

[crazylegs]

Originally posted by: Barfo

Originally posted by: Wreckage

Originally posted by: Scali
Hum, it would be a shame if the RV870 is indeed just an RV770 pimped up to DX11 spec. That seems to imply that there will not be a considerable performance leap in general, not much beyond what you'd expect from a regular die-shrink anyway.

They will get destroyed in the next round if the rumors are true. They already took a bad enough beating this round, I'm not sure if they could afford such a catastrophe.

I was hoping for something new and another round of competition.

And another thread goes down the drain, well done fanboy!

Only took him 18mins to embarass himself this time.

 

[lavaheadache]

Originally posted by: Wreckage


They will get destroyed in the next round if the rumors are true. They already took a bad enough beating this round, I'm not sure if they could afford such a catastrophe.

I was hoping for something new and another round of competition.

huh? I'm not sure if I follow you. Seems like you are implying Amd got whooped by Nvidia this round(which is down right wrong), yet then say you are looking forward to another round of competition, which implies the 2 companies were competetive to one another(which is truth)


I think it is blantantly obvious that almost any card you buy from this generation has an equal from the other company.

I think practically everyone in the world who follows graphics cards would say pretty much the same thing unless, ofcourse... one was prone to like a particular card just because it is made by a certain vendor., this is ill advised however.
Here at anandtech some members meet in secrecy scheming to perform outlandish attacks on such people and it could hazardous to ones reputation to be such person.

 

[Wreckage]

Originally posted by: lavaheadache

huh? I'm not sure if I follow you.

I'm basing it on market share. How else would you measure such a thing?

They also held the performance crown in both single and dual chip configurations.

Most review sites gave them the nod at each market segment as well (GTS250>4850, GTX260 (216) > 4870, GTX275 > 4890, etc.)

Here at anandtech some members meet in secrecy scheming to perform outlandish attacks on such people and it could hazardous to ones reputation to be such person.

Do go on.

 

[crazylegs]

Originally posted by: Wreckage

Originally posted by: lavaheadache

huh? I'm not sure if I follow you.

I'm basing it on market share. How else would you measure such a thing?

They also held the performance crown in both single and dual chip configurations.

Most review sites gave them the nod at each market segment as well (GTS250>4850, GTX260 (216) > 4870, GTX275 > 4890, etc.)

Here at anandtech some members meet in secrecy scheming to perform outlandish attacks on such people and it could hazardous to ones reputation to be such person.

Do go on.

How about on the fact that at any price point upto about £250 (more if you count x2 cards...) that both companies offer pretty much equally performing products, thus are obviously not being beaten, never mind taking a 'bad beatign this round' :S

With the majority of sensible people able to see that ATI/AMD is responsible for driving prices down this round and that they have a price advantage in the middle sector of the market.

No more replies from me, a famous quote comes to mind ''dont argue with a pig, u'll both get muddy but only one of you will enjoy it''

ps. look at my sig i run intel cd2 and nv gpu, out of my last 6 gpus 5 have been nv (x1950xt being the exception) BUT if i had to buy now it would be 4870/4890 1gb, lucky i dont game soo much now so can wait till the next gen dust has settled 🙂

 

[KingstonU]

Originally posted by: Wreckage
They also held the performance crown in both single and dual chip configurations.

Correct, no argument, but this represents 1% of the market, the money is not made here.

Originally posted by: Wreckage
Most review sites gave them the nod at each market segment as well (GTS250>4850, GTX260 (216) > 4870, GTX275 > 4890, etc.)

I was going to say the exact opposite, most review sites gave 4850>250 (aka 9800GTX+), 4870>GTX (216), 4890>GTX275, and ATI mops the floor at anything below 4850, which is the majority of the market and profits.

I will say that if it wasn't for how awesome the 8800GT was, we would not have had the pressure to developpe the awesome 4XXX that we see today, which will hopefully be enough pressure for Nvidia to so something outstanding with GT300.

I agree, with crazylegs, last post from me as well.

 

[Cookie Monster]

What is this most review sites your always on about? From what I gather, its been the other way around.

Our own AT thinks not.

That's the update from field. It's only been a short time since some of these parts debuted and the market is already compressed like it hasn't been in quite a while. For those who can afford it, buying a video card today will get you a lot for your money.

In any case, there are two standouts today: the Radeon HD 4850 and the Radeon HD 4890. These two are really terrific values.

The bottom line is, we as a consumer is getting some terrific value for the money spent on video cards. Anyway enough OT.

If RV870 isn't a brand new architecture, its interesting to know how RV770 can be slightly (or maybe not but the article seems to suggest so) modified to become DX11 compliant. Its all speculation at this point so we wont know just how much RV870 resembles that of RV770.

But I think all the spot light is at GT300. Brand new architecture, brand new process node and first gen DX11. nVIDIA is taking a rather different approach then what they have done traditionally ever since you know what. Hopefully it doesn't backfire on them.

 

[Scali]

Originally posted by: Cookie Monster
If RV870 isn't a brand new architecture, its interesting to know how RV770 can be slightly (or maybe not but the article seems to suggest so) modified to become DX11 compliant. Its all speculation at this point so we wont know just how much RV870 resembles that of RV770.

Well, from the hardware side, DX10 and DX11 aren't that different. The main differences are the new tesselation stage, and the new SM5.0. But AMD already had a tesselator in their DX10 hardware... so all they had to do was expand their shader instructionset from SM4.1 to SM5.0. That's not too big a deal, only a handful of instructions and features, some of which AMD may have already implemented anyway, for GPGPU and/or internal usage.

The other main features of DX11 are Compute Shader and better multithreading support. Those are both software-related. Both AMD and nVidia already had GPGPU-capable hardware, and Compute Shaders will also work on SM4.0 and SM4.1 hardware. So nothing new there.
Multithreading is also purely a driver-thing, doesn't need changes to the hardware.

 

[Creig]

Wreckage is just following in the footsteps of a long line of over-the-top Nvidia fans. For some reason, when there's an overly disruptive person in Video, that person is invariably someone promoting/defending Nvidia. And it doesn't just happen here. X-bit labs even wrote a piece regarding this phenomenon:

Note: this article came out before the ATI/AMD merger, so they're mentioned as separate companies.

FanATIcs vs. Nvidia Fanboys: Statistics

This page may offend some of our readers and forum posters. We admit. Sorry. But we honestly think that it makes sense to post this kind of thoughts here rather than encourage witch hunting, which have already begun in multiply forums.

Those who admire products by Advanced Micro Devices and Nvidia Corp. are a way more aggressive in the forums and against the journalists compared to those who prefer products by ATI Technologies and Intel Corp. This is an observation that has been made in several years. We do not know why this theory works, but we do believe that the people do not receive salaries from the aforementioned companies.

The most recent example: at least three readers pointed out, in a quite aggressive way, that some of the GeForce 7800-based graphics cards, certain models from certain manufacturers, supported dual-link DVI outputs to support those ?well-known? and ?wide-spread? 30? displays in a response to our claim that users of those high-end monitors should choose ATI Radeon X1000-based accelerators to drive those LCDs properly. No adorer of the Radeon products pointed out that there are some Radeon X1600- and X1300-based graphics cards from some manufacturers that feature passive cooling systems in a response to our allegations of noisy cooling systems on reference designs of the mentioned products.

Another example: X-bit labs has been (re-)using results obtained on Catalyst version 5.9 with Radeon X1000 support for all its graphics cards? reviews in October and November last year (there were exceptions made for Doom III and Quake 4 benchmarks conducted using Catalyst 5.10a drivers which brought significant performance improvements for these two games) with only two objections made in a very polite way be our readers. Once we engaged Catalyst 5.12 in a review dated the 8th of December, 2005 , we were requested in a very aggressive manner by numerous readers to use Nvidia ForceWare 81.95 drivers dated November 23 and updated on December 9, 2005.

We have received numerous accusations of bias when we reviewed flagship ATI Radeon X1800 XT and X1900 XTX series and none when we reviewed Nvidia GeForce 7800 GTX and 7800 GTX 512 products.

For the sake of the truth we have to admit that several weeks ago we received a very emotional complain from a man claiming that his Nvidia nForce4-based mainboard did not satisfy his needs and that officials do not want to respond to him in support forums. When asked to clarify the issues, the reader has vanished into oblivion.

We do not think that people who write us and make their points are getting paid by certain companies, even though such way of pressing on the media would be very efficient. We do like all of our commentators: luckily, the majority of them really make pretty interesting points. We encourage them to write and comment and we do not know why some behave the way they do. At the end of the day, everybody prefers products he or she likes for some reason and nobody should try to offend him or her because of that.

Wreckage can be extremely annoying with his blatantly obvious cherry-picking of benchmarks and truth-twisting, but since he's already received at least four one-week vacations and one two-week vacation in the last year or so, I imagine he'll be gone from our midst before too long. He'll probably end up in an ABT anti-AnandTech pity party thread with Rollo, Gstanfor and the rest of the pro-Nvidia club members who've been permabanned from here.



This is the second time in recent months you've made off-topic baiting comments. Three days off this time around. You shouldn't need another warning to be reminded of the forum rules.

AmberClad
Video Moderator

 

[BenSkywalker]

Brand new architecture, brand new process node and first gen DX11.

nV is alredy ramping several different parts on 40nm build process, just they aren't ones that are heading into the end consumer market. They haven't changed tick-tock, it has protected them too well for them to change that approach.

nVIDIA is taking a rather different approach then what they have done traditionally ever since you know what. Hopefully it doesn't backfire on them.

I don't think nV sees AMD as a long term threat at all, and if we take an honest analysis of this generation they probably have a good point. Right now nV is spending ~25% more on R&D then ATi is generating in total revenue, this despite what people seem to think is a slam dunk by ATi parts this generation. nV is outselling ATi better then 2:1 atm, no matter what all of the websites want to get behind or the loyalists to ATi. Speaking strictly from a market perspective, ATi got killed this round. Yes, they put pricing pressure on nV, even with that pricing pressure excluding R&D and a stock buyback charge nV would have had larger profits this quarter then ATi had revenue.

But I think all the spot light is at GT300.

And with the GT300 we are going to see a larger rift in approach then we have seen to date between AMD and nVidia. nV is going after GPGPU computing in a very serious way, that is the entire direction of their GPU strategy at this point. Their chips will be considerably larger then ATi's, they should draw considerably more power and be more expensive to make. They should also obliterate ATi's parts in GPGPU tasks. How important that is going to be to people and how much broader application support becomes is going to play heavily on how this round plays out for ATi. Yes, I meant to say ATi there. Right now, people that really care about GPGPU are pretty much all buying nV anyway, it isn't like there is a serious competition at that level which shouldn't be surprising as ATi has put at best a moderate effort into that level of their chips. Not saying that are right or wrong, just saying how it is.

If GPGPU really does take off, as nVidia has obviously been hoping for, in the GT300 timeframe ATi is going to be in an extremely bad position. If it doesn't, next gen is likely to go close to how this one has. nV will be pushing hard for broader PhysX adoption, it wouldn't shock me to see them get it up and running on OpenCL and/or CS simply as they are likely to have a huge performance advantage running it and having full support will make it much easier to get developers on board. AMD is going to focus on gaming performance as it seems fairly obvious based on the information they have released that is the only thing they are working on.

Is nV going to dedicate so much die space to GPGPU they will allow AMD to take an actual performance lead? I'm not talking bang for the buck, but an actual performance lead? While I think it is unlikely(raw chip size would tend to give a big edge to nV), I certainly wouldn't rule it out. The architecture nV is talking about is going to have massive xtor demands for GPGPU functionality, and a hefty chunk of those xtors are ones that could have improved game performance more if they were utilized differently. This happening with GPGPU not making any considerable headway would likely result in ATi ending up in a vastly superior market position then what they are today.

I think most people that have a reasonable understanding of the market and what the companies are trying to do see GT300 for what it really is- a preemptive strike against Larrabee. nVidia is trying to build the GPGPU market around their architecture and at this point in time it seems to be working. That could change in dramatic fashion as it is a market that is still in its infancy, but as of right now it seems they are doing everything they possibly can to create a very difficult market for Larrabee to enter into.

For ATi this generation, I think they really need to increase their value perception to customers. Obviously something on their end isn't working if they can dominate the market in terms of performance at a given price point and still lose marketshare. Being outsold by your competitor 2:1 in what has always been a performance oriented segment when you have better performance per dollar spent indicates something needs to change. Is it more marketing? Better bundles? More in game promotions? I'm honestly not sure, but they need to find a way to convince more people to actually buy their cards then their core fan base.

ATi seems to be marching down the road to complete their Fusion concept, at this point it seems they are planning on yielding the high end to nV and Intel altogether and focusing on some sort of CPU/GPU hybrid. Not saying that is what they are doing, but it seems to be the path they are taking. Based on what AMD has stated on the 5x00 parts they have a shockingly small increase in xtor count over the 4x00 parts. Given the move to 40nm this should make for very cheap chips when yields improve, they should also be able to clock fairly high given decent yields and will likely offer a solid performance improvement over the current generation, but nothing I see indicates they are focusing on the high end performance segment from any angle atm. Maybe the information they released was a smoke show to hide their high end parts? That is certainly a possibility, but 1Billion xtors given their current generation and how much CS will require for additional xtor versus current generation seems to indicate that ATi is focusing more on clock speeds the functional units to increase performance this generation. That is certainly a valid approach, and one that would allow them considerable flexibility in filling gaps in their lineup quickly with a clock adjusted/volt modified part. It doesn't give the appearance that they are serious about going head to head with nV for the performance crown this generation though. Luckily for them, nV is preoccupied with Larrabee and GPGPU so their gaming performance is likely to suffer due to it giving them their best shot at success, as long as GPGPU doesn't take off.

While I look forward to the next round of parts, I don't see either company giving us any serious effort in maximizing gaming performance. I think the parts will certainly be faster then what we have now, but I am not expecting anything major on that end. Given that gaming demands have not been going up for years, this isn't necessarily a horrible thing, Cryostasis is the only thing we have seen to date more demanding then Crysis after how long? Gaming graphics are at the point of massively diminishing returns, the upcoming shift in approach to the market that we will see over the next few years is going to change people's outlook on the market considerably. Not saying better or worse for any particular company, too many varriable to know if anyone is 'right' or if things will end up somewhere down the middle. Either way, going to be interesting to watch.

 

[Scali]

Originally posted by: Creig
Wreckage is just following in the footsteps of a long line of over-the-top Nvidia fans. For some reason, when there's an overly disruptive person in Video, that person is invariably someone promoting/defending Nvidia.

It seems to go the other way aswell now.
My first few posts on this forum happened to be related to PhysX, an nVidia technology. Now I have been supporting PhysX even before nVidia acquired it, because I am a firm believer in hardware-accelerated physics for the future of gaming. It's the technology that I'm interested in, not the brand that's on it. Ageia, nVidia, PhysX, NovodeX, all fine by me. If AMD and Intel come up with something even better, more power to them. As long as we get this technology firmly on the map.

Sadly I was mistaken for a pro-nVidia/pro-PhysX fanboy, and was approached rather aggressively by a few members, and even one moderator.

Just because SOME people supporting nVidia technology may be doing so because of blind bias doesn't mean that EVERYONE is just blindly biased towards nVidia. There IS some merit to their technology aswell, and some people may just genuinely like this technology. That has nothing to do with fanboyism or loyalty. But there is inevitably going to be some brand or other on whatever technology is on the market today. It's just a fact that most research is done by large commercial companies these days. They drive technology.

If AMD comes up with some cool new technology next, I may pick up one of their cards. If their DX11 cards are any good, I will probably get one. Still doesn't mean that I won't like PhysX after that.

 

[videogames101]

After ATI got pounded in 29xx/38xx, this generation was a time where ATI reestablished itself. Even with superior cards, this gen Nvidia outsold ATI on name alone.

It's my opinion that if ATI wins another the next gen, they will not have a stigma attached to them anymore and sales will be much more telling of which cards are actually better.

 

[OCGuy]

Originally posted by: Creig
*snip*

Holy massive thread derail!

 

[Scali]

Originally posted by: videogames101
It's my opinion that if ATI wins another the next gen, they will not have a stigma attached to them anymore and sales will be much more telling of which cards are actually better.

Well.. I wouldn't be too sure of that.
For years, there have been many games in the "The way it's meant to be played" program. When you start these games, you get something like "Best played on:" and a big nVidia logo on the screen.
I think this has made nVidia an incredibly strong brand among the army of more casual gamers. Much like how Intel Inside was such a strong brand even during the years that Pentium 4 was trailing behind Athlon in virtually every way. Sales barely suffered for Intel.