NVIDIA Kepler speculation thread

Better they should copy the turbo with PowerTune: 250W PT limit and ~1,2GHz Turbo which kicks in, when Tahiti is at low utilization thorough front-end limitation.
 
Mr. Baumann already pointed out that AMD prioritized timing over performance when launching Tahiti.

Hmm, let's wildly speculate here. Let us assume that i work for a bagel making company, and they have a competitor. After many back and forth, and after waiting for a long time to return to a position in which i can sell expensive bagels, after my competitor knocked me out from there, forcing me to pull all sorts of gymnastics around the fact that I'm forced to sell cheap bagels, I finally put out the most daring bagel in quite a while. It implements truly novel (for me) stuff in terms of mixture and whatnot. It's not as easy to bake. I'm quite confident in it and my evaluation of where my competitor would be for the coming bagel-cycle, so I'm quite daring in my exterior-facing message. However, I mis-estimate the competitor's actions. What do I do:

  1. admit the mistake in a very sporty fashion, tilt my hat to the other guys and so on and so forth;
  2. stick my head into the sand and sing "lalalalalala I can't hear you";
  3. make it seem as if my wonder-bagel, that was to humiliate the competition, was actually aimed at something else, conveniently being vague with regards to the state of a true, potentially competition humiliating, superwonder-bagel, thus keeping hope alive as time passes.
I would theorize that only one of those variants allows one to survive in a semi-public facing position in any corporation worth its salt. Note that this is hypothetical and isn't intended to say anything about Kepler and SI, it's just a humble suggestion to not exactly take at face value stuff that gets publicly said by companies in general.
 
I'm only 95% sure who's who in the bagel world, and even less what point you're trying to make. Can we stick to car analogies? ;)
 
http://fudzilla.com/home/item/26308-nvidia-gtx-680-pixellized-in-more-detail
fudzilla said:
It looks like our sources were spot on yesterday when we heard that the new GTX 680 will feature two 6-pin PCI-Express connectors and have a TDP of around 200W (190W to be precise). When combined with 1536 shader units, 2GB of GDDR5 memory clocked at 6GHz and, according to what we heard, a very high GPU clock, it comes as no surprise that Nvidia is claiming higher performance per Watt for its upcoming GTX 680.

This is a big win for Nvidia as we aren't used to hearing performance per Watt from Nvidia and now it appears that the tide has turned. The performance of the GTX 680 is somewhere around Radeon HD 7970 as it wins in some games and benchmarks and loses in others. Of course, these are all results that came from Nvidia so we'll hold our judgement until we see some reviews.

As noted before, the card is based on Nvidia 28nm GK104 Kepler chip that is around 300mm² in size.
The bad part of the story is that our sources were again spot on with the US $549 price tag...

That sounds pretty negative considering the source. Around the same speed as 7970 in the nvidia-selected benchmarks and drivers (IOW: slower), and nvidia feels the need to bang the perf/watt drum instead. (wonder if the 190W is tdp or just nvidia-tdp).
 
What about mindshare? Clearly Nvidia enjoyed the title for single GPU since X1900XTX, it seems like they are willing to share it for the first time in x years.

When it was released, the 5870 was substantially faster than the top single Nvidia GPU available at the time (GTX 285).
 
When it was released, the 5870 was substantially faster than the top single Nvidia GPU available at the time (GTX 285).

But in that case Nvidias response took the crown right back. This team it seems their response may tie at best (at least for a good while), a substantial difference.

It is kind of sad AMD only beat Nvidia's prior gen top dog by 10-15%, yet Nvidia couldn't beat that.
 
But in that case Nvidias response took the crown right back. This team it seems their response may tie at best (at least for a good while), a substantial difference.

It is kind of sad AMD only beat Nvidia's prior gen top dog by 10-15%, yet Nvidia couldn't beat that.

Or this time their response was always scheduled to come much later and they were surprised that their mid range chip actually was able to put up a fight and thus enables them to price it in the enthusiast territory for a punch in the gut for us.
 
But in that case Nvidias response took the crown right back. This team it seems their response may tie at best (at least for a good while), a substantial difference.

It is kind of sad AMD only beat Nvidia's prior gen top dog by 10-15%, yet Nvidia couldn't beat that.

That's because Nvidia have lucked into it, unexpectedly their performance chip is competing with AMD's enthusiast chip and we are going to pay the price. It's AMD's failure to make a competitive part which means Nvidia are under little to no pressure to get aggressive with pricing. If the GK104 was 10-15% behind the 7970 (like the GF114 was 10-15% behind the 6970) then we would be looking at a $349 launch of the GTX660Ti with the GTX680 to follow in June based on the GK110 at $599.

I don't blame Nvidia for taking advantage of AMD's failure to profiteer, but it still sucks for consumers that we are getting performance chips for enthusiast prices. Our only hope is for AMD to cut their prices aggressively to protect market share, but with the CPU division tripping over its own shoelaces every five minutes I don't see it.
 
That means that Tahiti is very close to (epic) Fail Number 2 like the original R600 was. With the difference that this time it's not late and it doesn't have a relatively cheap 8800GT to fight against. :devilish:


I see what he says but... how exactly would that happen? Are the guys at Apple blind not to see that Intel's graphics products are next to tragedy with image quality, drivers, etc? :LOL:

Haswell will essentially kill off the segment that Apple would have used, meaning this market is dead forever. It may sound dramatic, but this is the end of the mid-range GPU segment as a standalone part. This most lucrative slice of the market is now on its last legs.

If low and mid-range videocards segments will be gone forever, then what exactly future do nvidia have? Considering they lost consoles too. :oops: And with these uber-high high-end videocards prices, next to no one will buy them, excluding professional segments...

:???:
 
R600 was much more thoroughly shellacked back then.

The apparent scenario right now is two chips trading blows, and while there are some differences in die size and power consumption, they look to be smaller in comparison to what R600 faced.
 
That means that Tahiti is very close to (epic) Fail Number 2 like the original R600 was.
R600 was an epic fail, especially because of its shelf life that was less than 6 months. Tahiti is far from it. Once we see a die shot of Pitcairn, it should be relatively easy to get an estimate of the cost of DP on the size of the shaders. If it's on the order of 20% (as I expect), then that would bring a non-DP Tahiti in at a smaller or similar size of GK104. What's really seems to be happening this generation is that Nvidia is closing the area efficiency gap that's been there for years. With 2 players close to maximum efficiency, going forward, the one thing that will determine performance will not be technical competency but a game of marketing bluff: how to trade off die size vs. performance vs. features (DP/ECC or not) vs. what your competitor will do.
 
So the trend has gone from "Tahiti only XX% faster than Fermi" to "AMD ripping us off" to "Kepler, wow amazing performance from such a small chip" to "Kepler's price is ok" to "AMD sucks for not giving a more faster card" and now "Tahiti is the second coming of R600". I think I've seen it all.:D
 
So the trend has gone from "Tahiti only XX% faster than Fermi" to ''GCN is the the best desktop graphics architecture and physical implementation'' to "AMD ripping us off" to "Kepler, wow amazing performance from such a small chip" to "Kepler's price is ok" to "AMD sucks for not giving a more faster card" and now "Tahiti is the second coming of R600". I think I've seen it all.:D

fixed lol
 
I have this great idea for a website: write a whole bunch of nonsense breaking news about company XYZ. Then a bit later, I write more breaking news that my earlier breaking news is probably too hard for company XYZ because of whatever other nonsense reason. Then even later, I write that my first breaking news won't happen after all, because, wouldn't you know it, it was indeed too hard for company XYZ. And then I write "I told you so." And nobody will ever be able to prove me wrong.

Rinse, lather, repeat.
 
Luck is where preparation meets opportunity. ;)
Just like AMD with RV770 and RV870, you don't luck into a chip that has the similar performance for significantly smaller die size.

I dont consider 300 vs 350 all that relevant a die size difference. I also suspect there's some fanboy number fudging going on those Kepler measurements as I understand it's actually pretty hard to get an exact die measurement (you actually have to rip the die out, which nobody will ever do).

Still funny how die size is really important now when all those years AMD led it was irrelevant...
If anything I'd compliment AMD on their larger die, as I think it's a better idea to have the performance lead with a larger die, than not (especially when it's 300 vs 350, rather than 350 vs 500+ as typically AMD has been relative to Nvidia in the past). I've been wishing AMD would start pushing their die sizes up for a while personally. Although they didn't, at least this time it is larger than the competitor.

Bottom line this is the first time in a long time, ages honestly, Nvidia apparently wont have clear single GPU performance leadership, for an indefinite period of time, and that is unequivocally a step back for them.

And hanging your hat on an alleged GK110 isn't much at this point, we dont know if it exists, if it will ever come out, when, it's performance level, etc at this point. By that time AMD could well have their next gen ready for all we really know.

This idea that "this is really the mid part but it's so great it competes with AMD's sorry high end" is suspect at best imo. The facts are they called it 680, they're pricing it at 550, the die size is similar to Tahiti (just like the performance), everything suggests it is the best they could do for the foreseeable future.

It seems in a sense they copy AMD's "small die" strategy. They may be working on a bigger die, but they simply cant get those out the door fast enough and are ever later with them. GK104 was "only" 3 months late.
 
So the trend has gone from "Tahiti only XX% faster than Fermi" to "AMD ripping us off" to "Kepler, wow amazing performance from such a small chip" to "Kepler's price is ok" to "AMD sucks for not giving a more faster card" and now "Tahiti is the second coming of R600". I think I've seen it all.:D

I think you about summed it up :LOL:
 
Back
Top