Nvidia Kepler (6XX-series) discussion

[inux94]

Oh my god you stole Jesse Cox's computer. (ur sig)

Nice to see another fan around here

[adece]

Nvidia's silence is killing me. I plan on buying a *60 Ti or *70 when I travel to the US in may. I hope they have released those by then, if not I'm going for 7870 or 7950 and they would have lost my money (even though I prefer Nvidia)

[Adappy]

Nvidia's silence is killing me. I plan on buying a *60 Ti or *70 when I travel to the US in may. I hope they have released those by then, if not I'm going for 7870 or 7950 and they would have lost my money (even though I prefer Nvidia)

Yeah, it seems scary similar to how AMD was silent about Bulldozer's performance up until just before release... and we all saw how that release was . Personally, I prefer Nvidia over AMD, but if AMD is gonna have the performance crown for this generation then I may go with them again.

[kidsafe]

I believe price for the gk104 is supposed to be set at around $299

Everything we know about Kepler is a rumor including that $299 pricepoint for GK104. It is reasonable to expect any GPU Nvidia puts out for $299 to be roughly equivalent in performance to a GTX 570 of Radeon HD 6950. If Kepler really does devote transistor space to PhysX processing, then its outright performance in the vast majority of games might be even lower than expected.

[Deleted]

The only problem I see with releasing all these cards that are X amount of times better than the 580 or something about blowing the 7970 out of the water.. is that most people right now dont even NEED a single 580 to max out 99% of games out in the market. The big problem is that we are still paying the same amount now as we did 1.5 years ago for the same performance. For example, the 6950 was $259 at launch.. and honestly, that about the medium price now.

I can see a card coming out that will replace the 560, lets say, but it will cost more, for more performance. Doesn't make sense to deviate from that 560. And those ultra high end cards.. well they are going to simply RAISE the money cap, instead of pushing everything down. In other words, lets say the next greatest card comes out and costs $500. Then all the other, more mainstream cards will come out lower and lower than that, which means we get plenty of price wars as consumers and more value for our money.

Now, instead, the high end cards will continue to cost more and more.. I can see the 680 coming out costing $600, which means that prices at the mainstream level remain the same for roughly the same performance.

can't wait for new series to come out so the old cards drop their prices so i can get a 580/590

[nbm02ss]

can't wait for new series to come out so the old cards drop their prices so i can get a 580/590

Good luck finding a 590, and I really mean it. Newegg has them as deactivated and the ones on Amazon are suffering from some major price gouging. =\ By the time Kepler is out I doubt if you'll be able to find a new one.

Looking forward to what the 690/790 will be capable of though.

[Rennadrel]

Yeah, it seems scary similar to how AMD was silent about Bulldozer's performance up until just before release... and we all saw how that release was . Personally, I prefer Nvidia over AMD, but if AMD is gonna have the performance crown for this generation then I may go with them again.

Price should be as much of a factor as performance. If you are an enthusiast with a big budget and can spend as much as you want on a graphics card or two, fill your boots and buy the best performance for the dollar, however I think for the general consumer, price and performance go hand in hand. I take that approach with graphics cards and prefer AMD for this reason, the last two GPU generations they have put out were a lot closer to Nvidia's offerings then in years past, and at a fraction of the price. However with AMD aiming for a higher end enthusiast market with higher priced top end cards then in the past, I have to weigh my options and buy the best bang for my buck that either company is going to offer me.

[Deleted]

More than happy to wait and see what they bring out i would rather them take time getting it out than it be rushed and messy on release.

[adece]

Thought this thread was worth bumping for this.

Nvidia Struggling With Poor 28nm Yields

[Rennadrel]

Thought this thread was worth bumping for this.

Nvidia Struggling With Poor 28nm Yields

Pretty sure this was known back in December was it not? I think AMD had started their development of the 28nm cards earlier and were able to yield quicker results just due to the overall development time advantage.

[kidsafe]

Pretty sure this was known back in December was it not? I think AMD had started their development of the 28nm cards earlier and were able to yield quicker results just due to the overall development time advantage.

It wouldn't be surprising since GF100 was delayed for 6 months and basically flopped. This forced Nvidia to quickly spin up GF104 as a stop gap in the midrange while still needing GF110 and GF114 later. That's a lot of wasted time by their dev team when AMD only really spent significant time designing the HD 6900 (Cayman) GPUs.

More pressing is the assumption that GK110 is massive and GK104 is roughly the same size as the Radeon HD 7900 (Tahiti). Just the design of such a behemoth is complicated enough, let alone the manufacturing/lithography on TSMC's side. As all 28mm TSMC customers are paying per wafer regardless of yields/defects, this results in Nvidia having to spend the same amount for lower yields altogether and higher percentage defects.

[Nab]

I'm pretty sure if they decide to delay it then it will be significantly better than the 7*** series.

[DeltrusDisc]

Gotta agree Synthaxx, it's like AMD's Bulldozer all over again, so many AMD people after so many months of waiting ultimately bought an Intel Sandy Bridge and only the most hardcore of AMD fans stuck it out, and were sorely disappointed, of course.

---------- Post added 2012-02-20 at 04:54 PM ----------

I'm pretty sure if they decide to delay it then it will be significantly better than the 7*** series.

Read what I just wrote. AMD delayed the Bulldozer chips like half a year until finally releasing them.... and they sucked in comparison to the Intel Sandy Bridge, which had been out for about 10 months. Your logic is pretty flawed. >_<

[Nab]

Gotta agree Synthaxx, it's like AMD's Bulldozer all over again, so many AMD people after so many months of waiting ultimately bought an Intel Sandy Bridge and only the most hardcore of AMD fans stuck it out, and were sorely disappointed, of course.

---------- Post added 2012-02-20 at 04:54 PM ----------



Read what I just wrote. AMD delayed the Bulldozer chips like half a year until finally releasing them.... and they sucked in comparison to the Intel Sandy Bridge, which had been out for about 10 months. Your logic is pretty flawed. >_<

Yes, but this is Nvidia we are talking about. They usually come with surprises.

[BicycleMafioso]

I'm pretty sure if they decide to delay it then it will be significantly better than the 7*** series.

Not necessarily. Look at Fermi and their massive delay compared to the HD5000-series.

Yes, they were a bit better, but at what costs? +60-70~% price, +100%~ more power (a rig with two HD5870 drew less power than one GTX 480), +25% more heat and +10-15% more performance.
And that took them 6~months longer than expected.

---------- Post added 2012-02-20 at 05:57 PM ----------

Yes, but this is Nvidia we are talking about. They usually come with surprises.

It is indeed nVidia. I am hesitant to that meaning they are better.
They are by no means more of a holy grail.

[kidsafe]

I'm pretty sure if they decide to delay it then it will be significantly better than the 7*** series.

I wouldn't be so sure. GF100/Fermi was 6 months late, too expensive, too hot, and barely 15% faster than a Radeon HD 5870.

[Deleted]

I think the thing that is worth noting the most is after AMD release the flagship 7990 (that will most likely be before the first keplers too) they will be already talking up the next gen/die shrink 8000 series, All this even before Kepler is released. In development terms AMD is ahead of Nvidia as they have a 6 month cushion and the only way Nvidia will get over it is if the GTX600 series is a serious increase in performance over the 7000 series. Performance alone won't even be enough they will have to be as cool and as efficient too and the right price!

Tough times ahead for Nvidia they might not claw it back until next year now.

[BicycleMafioso]

I think the thing that is worth noting the most is after AMD release the flagship 7990 (that will most likely be before the first keplers too) they will be already talking up the next gen/die shrink 8000 series, All this even before Kepler is released. In development terms AMD is ahead of Nvidia as they have a 6 month cushion and the only way Nvidia will get over it is if the GTX600 series is a serious increase in performance over the 7000 series. Performance alone won't even be enough they will have to be as cool and as efficient too and the right price!

Tough times ahead for Nvidia they might not claw it back until next year now.

Which is quite ironic, seeing in AMDs other field, they are starting to become years after intel. Which is a sad thing.
A good thing they've realised, though, and shifts their interests from high-end CPUs and toward a stronger integrated GPU though. A shame for workstations and enthusiasts, but it's what will likely win them big shares of the market in the long run.

However, I think they should still have an architectual fine-tuning of 28nm(?) before they shrink again.

[kidsafe]

I would not expect 20nm half-node from GlobalFoundries or TSMC until 2H 2013.

[Deleted]

http://vr-zone.com/articles/report-n...-ti/14952.html

Nothing special, if the performance part is true. But it will depend on the pricing ofcourse