NVIDIA reveals Fermi's successor: Kepler at 28nm in 2011, Maxwell in 2

[Mousemonkey]

They made announcements about inventory as well. You can read many things in to that. Sometimes its reporting manipulation other times it means they are selling well.

Well they haven't sold anything to me recently and won't do until they release a fully enabled GPU of 460 class or higher.

 

[notty22]

I just ordered my second 460
Will be asking all sorts of stupid sli questions, if my favorite games don't utilize.
But I'm reading, like magic , most will ?

 

[Mousemonkey]

I just ordered my second 460
Will be asking all sorts of stupid sli questions, if my favorite games don't utilize.
But I'm reading, like magic , most will ?

For the ones that don't support it natively, forcing AFR or SFR via the Nv control panel is always an option.

 

[JAYDEEJOHN]

Yep, the future looks good

 

[ares1214]

:lol: Yeah, i was kinda wondering why they were doing it on a per watt basis, fermi isnt exactly the greenest video card

 

[JAYDEEJOHN]

Well, I believe using a smaller process, a more mature core,HKMG and some tweaks, plus making them Cuda cores more responsive etc, theyll hit their goal.
How that effects gaming and gamers?
Yet to be seen

 

[ares1214]

Right, but CUDA is likely the thing NV can twist the most. Not saying it isnt useful, not saying it doesnt contribute to performance, but its like AMD bragging they have 4x the shaders than NV. Then making a chart showing that. Also, if you think about it, if they cut the power to 5870 levels, they already have a 30% decrease in wattage, so id rather see raw performance, not performance per watt, and what might be idle watt or something misleading. I never trust AMD or NV slides when it comes to performance. They can skew the results, and market it however they want.

 

[nforce4max]

:lol: Yeah, i was kinda wondering why they were doing it on a per watt basis, fermi isnt exactly the greenest video card

Neither was this...

http://www.youtube.com/watch?v=IFcJvNAXi7w

 

[JAYDEEJOHN]

Honestly? I feel like this guy

 

[ares1214]

HA, both of them :lol: I wonder if NV imagined that powered the fermi? :lol: And believe me, i want NV to get back in the game for sheer competition. But at BEST, we will see this Q3 2011, in which time 6xxx has already had 8-12 months on the market, and 7xxx is right around the corner. I also dont imagine many refreshes on fermi until then, as they have it clocked as high as they basically can, clocking it higher would just make it run hotter and use more energy. They cant shrink the die, not for a LONG time, and thats what this is. GF104 can only take them to maybe around 480 + or - 5% performance as far as single cards go. Its already taking them long enough as is to finish off Fermi. Their relations with TSMC, while they may be strong, might be strained, as they are a decent bit of the problem, and NV has been talking with GloFlo for what must be Tegra. I dont see them doing well for a while, but thats just me. Maybe another player can help lower AMD pricing.

 

[notty22]

Nvidia's position might get better with the release of the AMD 6 series, it may FAIL.
The only things we know for sure, ARE, its still on 40nm, and every gpu should be a bigger die. Which will probably use more power, and could run hotter
Time will tell.

 

[nforce4max]

I hope the early 6k series does fail and it does look disappointing for those who already own a decent 5k series or GTX 4xx. To be honest I think ATI is only trying to milk the industry while it can. Like what Nvidia has done in the past but without competition it doesn't matter what brand that is top hog it is us the consumer that suffers with needlessly higher prices. I still remember the court case ware the both of them were fined for making agreements to fix the prices of each other products to keep them artificially high.

 

[ares1214]

I hope the 6xxx series does good. One should never hope someone does bad to bring more competition in. Best case scenario is 6xxx does good, which it really does look like it will, if they keep temps in a good range, and then they price to kill :lol:

 

[JAYDEEJOHN]

If the 6xxx series is only better than the 5xxx series, like nVidias 8xxx series to the 9xxx series, then I'll agree
We shall see.
I still think some form of a full chip could be done, and nVidia will have the time for it, as 28nm isnt going to be here anytime soon

 

[Mousemonkey]

I think the Tessellation performance is going to be one of the main areas that they may concentrate on as the performance so far has been a bit lacklustre considering the noise made about being able to do Tess since the 2xxx series and how it was held back because of Nvidia's refusal to bow down to DX10.1.

 

[JAYDEEJOHN]

I give alot of credit to nVidia for having a nice first DX11 gen card, usually its always the second gen
Yep, ATI has some catching up to do there, where nVidia just needs to get everything out of Fermi, but Im thinking that ATI too may be trying to do the same with their cards as well, not just tess
Their flops dont measure up to perf, so theres alot of wiggle room there, sorta like Fermi as well
Theyre both missing pieces, and when they fix them, we will all be happy

 

[Mousemonkey]

But it's the tess chip that either needs a redesign or dropping and incorporating the process into the GPU that could be the make or break I reckon, not that I'm claiming to know the full ins and and out of it all but that seems to be one of the areas that they are getting hurt in and it just so happens to be a selling point of DX11.

 

[JAYDEEJOHN]

Their second gen is coming soon to a rig near you

 

[ares1214]

But it's the tess chip that either needs a redesign or dropping and incorporating the process into the GPU that could be the make or break I reckon, not that I'm claiming to know the full ins and and out of it all but that seems to be one of the areas that they are getting hurt in and it just so happens to be a selling point of DX11.

6xxx has an entirely new tess system. Upside is on the high end models, it gets fantastic tes performance. Downside is, the more shader banks you take away, tess gets hurt a lot more than it should be.

 

[Mousemonkey]

6xxx has an entirely new tess system. Upside is on the high end models, it gets fantastic tes performance. Downside is, the more shader banks you take away, tess gets hurt a lot more than it should be.

And how do you know this? We have yet to see either the cards or independent benchies.

 

[EXT64]

He didn't say tess, he said tes, which refers to The Elder Scrolls, aka Oblivion, which it likely (as do all modern cards) does quite well in.

 

[saint19]

+1 to mousemonkey, AMD 6xxx and Kepler hasn't been released even hasn't been confirmed until now are just rumors and as rumors we only can discuss the possible release of both :lol:

Just my two cents

 

[4745454b]

So who has 28nm coming out in the time frame that Nvidia will need to do this in 2011? As I recall thats one reason why AMD had the "isle" problem and changed what the 6series will be. 6 series is more of a refresh then a new card, and I'm not sure Nvidia needs to do much about it. The GTX4xx cards are faster then the AMD 5xxx cards, so let them refresh it. Even if AMD can get the lead back, when the "real" new cards come out we'll be back to seeing who is fastest again. (7 series and GTX5 series.)

 

[ares1214]

And how do you know this? We have yet to see either the cards or independent benchies.

Know some people, see the arch, and that one leak. Leak may or may not be real though. Its just how its made differently in the arch. It wont scale linearly between high end and low end. Dont believe me if you dont want to, but there it is.

 

[Griffolion]

The performance per watt graph isn't very encouraging because it doesn't necessarily mean that Kepler will be massively faster than Fermi, just a lot more efficient!

We can only wait and see in any case, the whole rumour/speculation/armchair engineer thing is always fun!