nVidia's 28nm Kepler expected to hit stores in late 2011?

[G@merX]

Not sure why I'm considered a troll for having an opinion and placing it on a forum? The nerve of me..

[Roadhog]

Originally Posted by G@merX

Not sure why I'm considered a troll for having an opinion and placing it on a forum? The nerve of me..

Usually what happens when you spout lies like ATI has poor IQ.

[Rollo]

Originally Posted by Redeemed

Wow- I think that speaks volumes of where nVidia is. That they're no longer just a GPU company. I can't believe that they'd go that long without a competitive release. They're practically handing the victory to AMD by delaying their parts for that long. Going off of that chart AMD will have close to 8 months with zero competition in the low end and mainstream, and up to 6 months with nothing competing with their high-end. I suspect GK112 will out perform the Radeon 7990 or whatever but will that be enough to off set the 7990 being available 6 months earlier?

I just get the impression nVidia isn't worried about remaining uber-competitive in the high-end discrete GPU market. They've got their focus else where is what seems to be happening.

No doubt! Did you hear about how they fired their lead GPU designer, and the whole GPU PR and marketing team?

Oops...my bad! That was ATi.....

[shadow001]

Originally Posted by Rollo

No doubt! Did you hear about how they fired their lead GPU designer, and the whole GPU PR and marketing team?

Oops...my bad! That was ATi.....

It wasn't their Lead GPU designer, it was the product manager that pushed to add eyefinity as a feature on the HD5000 release, wich of course Nvidia has yet to implement it while using a single card now that we're at it....

[Rollo]

Originally Posted by shadow001

It wasn't their Lead GPU designer, it was the product manager that pushed to add eyefinity as a feature on the HD5000 release, wich of course Nvidia has yet to implement it while using a single card now that we're at it....

Personally I don't care if NVIDIA ever implements surround on a single GPU. There aren't any GPUs powerful enough to do it well.

Now a better question would be:
1. When will ATi give us the GPU accelerated physics they've been promising for years and years and years?
2. When will ATi give us multi GPU 3d?
3. When will ATi give us multi screen 3d?
4. When will ATi step up to the plate with developer relations and have multi GPU profiles ready for most big games pre-launch?

[i SPY]

Bullet is on the scene since end 2008, although its not so widespread yet (3dmark2011, Gta4, Lanoire, Maxpayne3 and few more), but its at least 1000x more efficient then any physx 2.x.x *cough crap cough* code. Even Physx 3.x.x is questionable at its best.


As for the rest, imo its still irrelevant and poor gimmick to spend more money. True 3d will be when you dont need glasses for it - let display do 3d and gpu for what its meant to do render polygons, post process, etc instead of wasting extra cycles for this "3D".

[Rollo]

Originally Posted by i SPY

Bullet is on the scene since end 2008, although its not so widespread yet (3dmark2011, Gta4, Lanoire, Maxpayne3 and few more), but its at least 1000x more efficient then any physx 2.x.x *cough crap cough* code. Even Physx 3.x.x is questionable at its best.


As for the rest, imo its still irrelevant and poor gimmick to spend more money. True 3d will be when you dont need glasses for it - let display do 3d and gpu for what its meant to do render polygons, post process, etc instead of wasting extra cycles for this "3D".

ATi doesn't share your lack of concern. they've been trying to do this stuff for years! (although it could well be ATi's new CEO doesn't care about enthusiast tech like the things I mentioned either)

[Razor1]

Originally Posted by Ninja Prime

Sorry, I was already irritated by someone else, I might have been a bit of an ass.

Reasoning being, 28nm doesn't have the performance scaling to double performance. Leakage is getting worse, this is why Intel switched to 3-d gates at 22nm, performance characteristics are getting worse and worse for each shrink, for traditionial processes. For example, from 4870->5870 doubled and in some cases more than doubled resources, more than doubled transistor count, from 953 million to 2.15 billion, and increased clocks. Yet performance scaling was only generally in the 60-70% range. If even if we expect kelper to be over 6 billion transistors, as massive or more massive than fermi was(which I don't) and thus double resources, its not going to get 100% scaling. Also, at best, they can only get ~50% more bandwidth unless they go back to 512 bit bus, which makes it even bigger, maybe GTX 280 size. Even that 50% is going cost them a big chunk fo die size for a better memory controller. Reportedly the memory controller from 4870 -5870 had to double in size to handle the faster clocks, also reportedly this is the reason for fermi memory controller not being able to scale up very well beyond 4ghz, because its not beefy enough.

I don't see any way they can make kepler beat a 590 unless its in some benchmark that previously had some odd bottleneck that they've fixed, ala Heaven bench. The hard info we have is that DP is "over 1 Tflop" on kelper, which while its not specific, it kind of gives us a ballpark. Thats only a 25-30% increase. If it was closer to 1.5, or so which would be close to double performance, they probably would have bragged up "over 1.5 Tflop" right? Seems to me that they aren't going for such a huge chip this time, and are trying more for efficency per die area.

Edit: On the "two different architectures" thing, reports are kelper is mostly a die shrink of fermi without too much new. /shrug

Its a different architecture, well heavily modified and its supposed to have close to 3 times the DP flops as Fermi. And DP gives us no hint on actual performance in all tasks outside of things that need DP.

Leakage that is true, but it will depend also on the architecture of the chip and maturity of the process.

[shadow001]

Originally Posted by Rollo

Personally I don't care if NVIDIA ever implements surround on a single GPU. There aren't any GPUs powerful enough to do it well.

Now a better question would be:
1. When will ATi give us the GPU accelerated physics they've been promising for years and years and years?
2. When will ATi give us multi GPU 3d?
3. When will ATi give us multi screen 3d?
4. When will ATi step up to the plate with developer relations and have multi GPU profiles ready for most big games pre-launch?

That depends on wich games you play to allow 3D screens with 1 card and remain playable, and the hardware gets faster and faster over time anyhow, with AMD's next big launch in mid january, and looking promising btw(possibily close to the speed of an HD6990 in a single GPU card), and the dual GPU variant of this monster in march.....So where is kepler?


As for physics, you got your answer with bullet and not being tied to a specific vendor being it's main atraction to developers....They like open standards that work on every card.


As for 3D viewing, it seems Samsung showed a prototype 27" LCD where it has a built in camera on it's outer frame that's precise enough to follow eye movement and send different data to each eye, wich means not wearing any dorky looking 3D glasses anymore( they are ugly ****ers even in version 2.0).


As for using 3D viewing across 3 LCD screens, i don't think that's even possible unless you grow a third eyeball so that specific information is sent from each specific display to each eyeball, so the system has to be radically different and not based on current tech at all( holographic display perhaps?) .

[Rollo]

Originally Posted by shadow001

That depends on wich games you play to allow 3D screens with 1 card and remain playable, and the hardware gets faster and faster over time anyhow, with AMD's next big launch in mid january, and looking promising btw(possibily close to the speed of an HD6990 in a single GPU card), and the dual GPU variant of this monster in march.....So where is kepler?

We were talking about single GPU surround, and unless you have the very highest ATi GPU it would seem pointless to me. (and limited at the highest) I say this as a person who's played lots of 1080p surround on dual 560Ti 2GB cards- a considerably more potent rig than a 6970. Beats me where Kepler is, but we have no valid launch date for next gen ATi either.

Originally Posted by shadow001

As for physics, you got your answer with bullet and not being tied to a specific vendor being it's main atraction to developers....They like open standards that work on every card.

Devs like open standards? You listed a couple Bullet games. I have around 15 PhysX games. I guess devs prefer physX, and likely will even more with the version 3 toolkit.

Originally Posted by shadow001

As for 3D viewing, it seems Samsung showed a prototype 27" LCD where it has a built in camera on it's outer frame that's precise enough to follow eye movement and send different data to each eye, wich means not wearing any dorky looking 3D glasses anymore( they are ugly ****ers even in version 2.0).

Oh, a prototype at a tech show. So like I said, ATi doesn't really have any competitng 3d, you're just trying to say the 3d of the future will be better? OK, I'll buy that. The CPUs, cars, cameras, stereos, every thing else of the future will likely be better too.

Originally Posted by shadow001

As for using 3D viewing across 3 LCD screens, i don't think that's even possible unless you grow a third eyeball so that specific information is sent from each specific display to each eyeball, so the system has to be radically different and not based on current tech at all( holographic display perhaps?) .

You should come over to my house and try it. I do it every week, and like it so much I bought 5 more 3d monitors (NVIDIA gave me one) to do just that.

http://www.tomshardware.com/reviews/...g,2672-10.html

In games that are “3D Vision-ready,” the experience is as profound as going from a 2D VGA card to a 3D graphics accelerator. I BS you not. If this technology didn’t have such a steep barrier to entry in its cost, I can guarantee you’d be all over it.

You know what reviewers say about ATi's 3d?

ATi 3d sucks.

[i SPY]

Devs like open standards? You listed a couple Bullet games. I have around 15 PhysX games. I guess devs prefer physX, and likely will even more with the version 3 toolkit.

Fanboying much?

Physx 3 could bring a change but not much, bullet is still more efficient.


There are like 4-5 HW physx games worth mentioning, rest is just physx milking.. And even these 4-5 newer games (Alice2, Mafia2, BatmanAA, AC and maybe Cryostasis) are all bugged/inefficient.. Same story with a dedicated gpu.


For example Alice2 with 60fps cap, physx - high;
near fluids and smoke it drops to 35fps >> that's with gpu usage at max 52% (570gtx).. yes great efficiency, same sheit in batman AC.



Btw speaking of open std., Nvidia is thinking of importing Physx to openCL. Im kinda surprised you never heard of it since you're such a nvidia "expert".

[Rollo]

Originally Posted by i SPY

Fanboying much?

Physx 3 could bring a change but not much, bullet is still more efficient.


There are like 4-5 HW physx games worth mentioning, rest is just physx milking.. And even these 4-5 newer games (Alice2, Mafia2, BatmanAA, AC and maybe Cryostasis) are all bugged/inefficient.. Same story with a dedicated gpu.

For example Alice2 with 60fps cap, physx - high;
near fluids and smoke it drops to 35fps >> that's with gpu usage at max 52% (570gtx).. yes great efficiency, same sheit in batman AC.

The question was which standard developers prefer, not which is more efficient. Only two games currently released were given as examples of Bullet, I have around 15 PhysX games on my shelf. It's pretty obvious which developers prefer.

As far as "ye olde but some of those games are bad games!" argument goes, so what? The question is which developers prefer and obviously it's PhysX. I could say the same about Direct3d- 90% of the games aren't what I'd call "good games". Doesn't mean there's anything wrong with Direct3d.

Originally Posted by i SPY

Btw speaking of open std., Nvidia is thinking of importing Physx to openCL. Im kinda surprised you never heard of it since you're such a nvidia "expert".

Source? Here they are a couple months ago not only saying they have no plans to do it, but that developers aren't even asking them to do it:

http://physxinfo.com/news/6419/exclu...sx-technology/

Is the fact that GPU PhysX is exclusive to NVIDIA some kind of barrier for GPU PhysX adoption?

Rev Lebaredian: It is not a secret that most of game developers are concerned about consoles. PC for them is a smaller SKU usually, so PhysX not running on competitor GPUs does not really matter that much to them.

Tony Tamasi: For every title we have worked on; for every one that I’m aware, that was never a reason for someone to use or not to use GPU PhysX.

PhysXInfo.com: Do you have any plans to port GPU PhysX to DX11 (Direct Compute) or OpenCL?

Rev Lebaredian: No plans to do it. Maybe if there will be a good reason for that in the future, but there is no current plan.

PhysXInfo.com: So developers are not asking you to port GPU PhysX to OpenCL?

Tony Tamasi: Nope.

Rev Lebaredian: No.