What if 1080Ti never arrives?

Cuz we hAve the specs and almost anyone can come up with average gains over 1080. Not rocket science.

Sure, we've had the specs for lots of cards over the years.

And there's always someone thinking they're going to apply simple math and get a simple solution.

And it usually turns out that it simply isn't the case.
 
Sure, we've had the specs for lots of cards over the years.

And there's always someone thinking they're going to apply simple math and get a simple solution.

And it usually turns out that it simply isn't the case.

Not when it comes to same architecture. Titan x is based on pascal if it was on newer architecture then yea. 60% over previous Titian? When 1080 is already 30% or so if we took Nivida at their word and gave them 100% benefit of the doubt. We are looking at 30% faster than 1080 at best.
 
Since AMD has bowed out of yet another round, why shouldn't they simply go after that tall enthusiast cash?
I think the problem here is with the prices where they are at a 1080Ti would be to close to a 1080 and the titan in price performance. I dont see it happening right now.
 
Why would Nvidia be in such a hurry to compete with itself?

For the same reason Intel is. AMD have no influence in the segment anymore. But you still want to sell people new stuff to make money. Innovation and progress is secured without competition in a business where cash flow and volume is vital.
 
Would Nvidia really be competing with itself? They just need to keep the cost difference 1080<1080Ti<Titan X. Not sure why they would need 3 tiers of high end cards...but they need to make the 1080 Ti so I can buy 2. That's all.
 
The only way the 1080ti doesn't appear is:
  • They decide to call it a 1180 or a 1280 or whatever instead (and what's in a name, anyways, if it is a Pascal GP102 variant)
  • AMD decides to pack it in
  • Time grinds to a halt (hey, you never know..)
 
$600 1080, vs a $1200 Titan. NV is going to need a $900 part... My guess is if the 1080 Ti comes in a few months, it'll be a ~$950 3200 core part. I spent $1000 on a GPU once a before (my two extra Titans cost me $600 and $400), and I'll never do that again. So realistically the 1080 is as high as I will consider this generation.
 
Would Nvidia really be competing with itself? They just need to keep the cost difference 1080<1080Ti<Titan X. Not sure why they would need 3 tiers of high end cards...but they need to make the 1080 Ti so I can buy 2. That's all.

Yea... Nvidia likes to keep something up their sleeve...which they can, considering the market.
 
$600 1080, vs a $1200 Titan. NV is going to need a $900 part... My guess is if the 1080 Ti comes in a few months, it'll be a ~$950 3200 core part. I spent $1000 on a GPU once a before (my two extra Titans cost me $600 and $400), and I'll never do that again. So realistically the 1080 is as high as I will consider this generation.

I think we could see the Ti having same core count and RAM as the Titan this time. Depends how good yields are. If it's same specs, they'll have an easier time pushing that higher price point.

If this happens they will just refresh the Titan with 24GB RAM and full 3840 core count, probably under a new name too. This new Titan X will be quickly forgotten then, there are many people who don't follow GPU launches closely forget they launched a cut down Titan with Kepler as well. They have an opportunity to do so this time as unlike the original X, it is a slightly cut down chip.
 
The Titan cards always come out first.

You can bet that the Pascal Titan will be MUCH more than $999

The 1080 already is over that price here in Australia. At $1149.

The Titan X will cost close to 2k here once its released...
 
This whole Pascal thing is on its way to a shitstorm it seems. Sure the cards are faster than Maxwell, but something is brewing.
Complete lineup release in a few months and now rumors of Volta coming in 2017 (on 16nm) must mean something. I am a bit excited :)
 
This whole Pascal thing is on its way to a shitstorm it seems. Sure the cards are faster than Maxwell, but something is brewing.
Complete lineup release in a few months and now rumors of Volta coming in 2017 (on 16nm) must mean something. I am a bit excited :)
They make new cards to an approximate schedule and release them.
Seems quite usual.

Whats this shitstorm you speak of?
 
No need to panic until we get reliable info about volta. I have very huge doubts about anything volta-related being released for consumer market before 2018. It might just be a 750ti like card using newer architecture. Anyway, there's still a lot of time to enjoy those 1080s before starting to post on craigslist (and here i was hoping to not bother for a few years lol. With rumors like this...)
 
I don't think we'll see a full 24GB Titan. My guess now is that an $799/899 1080Ti would have to be a near 12GB Titan XP clone less 256 cores, and then Titan Black XP gets a refresh to 3840 cores but still 12GB. That scenario would have to be a ways off in the future, like maybe Feb-Mar 2017, 8-9 months from now.
 
Tell me why in the hell would anyone need an absurd 12GB of ram on a video card? Games would never ever come close to using that much ram. Nvidia is getting out of hand. Titan X is for suckers!
 
Whats this shitstorm you speak of?

NVIDIA Announces Quadro Pascal Family: Quadro P6000 & P5000

This confirms a 3840 Core GP102, Quadro P6000, due in October.

Seems Nvidia is just selling the chips that don't make the cut as the new Titan X. Probably means we'll see another Pascal Titan a la Titan Black at some point when yields get better, as some have predicted.

That is what I'm talking about ;)
People should avoid Pascal. There probably will not be a 1080Ti. This new Titan X is what 1080Ti should have been, and Quadro P6000 is what new Titan Pascal should have been.
Now there are rumors that Volta is coming next year, on 16nm, so not true Volta, but still, This new $1200 Titan X might be beaten in a year by a card that's $450... and if AMD somehow actually manages to get in the same performance as 1080 or maybe even Titan X then nVidia will definitely launch something even faster (and/or probably cheaper).
There are just too many factors to actually spend $1200 on 1080Ti disguised as Titan which could become near obsolete in a short time. Buying Pascal just doesn't seem like a sensible decision.
 
Last edited:
Depends what card you already have, how much you need extra performance and what you can afford.
If they dont release a 1080ti, they lose lots of sales. Not a shitstorm though, the cards released are pretty good.
I would wait until iron clad info is available on the ti before melting down.
Even then, consider staying solid :p
 
This thread title reads like a 10 year-old crying about "what if Santa never comes, momma"? There will be, without a gnat fart's shadow of a doubt, something to fill the $700-1200 gap coming from nV.

If I was nV, I would be laughing at a thread like this since it will push people who are waiting to buy 1080 ti. This is exactly what they want, and the reason they time their "ti" models to come out just as interest is waning (and prices start reducing/stabilizing on their flagship). In fact, if I was nV marketing I would be directing my outsourced astro-turf marketing drones to make threads like this :)
 
Given all of this, worth purchasing 1080s in SLi or just keep the 980 Ti SLi setup until November?
 
Paxwell wasn't even on the roadmap in 2014. It's a rushed response to what they assumed was coming in Polaris form. The lackluster 1060 and average DX12 performance seem to cement this.

Volta is to compete with Vega.

Guess what Navi will likely have - SSG... 'next gen memory'

Roadmap-640x360.jpg
 
Given that there is no architecture named Paxwell, it's still not on the roadmap.
At this rate, Vega may not compete with Pascal.

Never heard of the term Paxwell? You must be new around here.

Who knows what Vega will do. It may compete with Volta for all we know.

What we do know is Paxwell was not so well planned.

OldRoadmap.jpg
 
Never heard of the term Paxwell? You must be new around here.

Who knows what Vega will do. It may compete with Volta for all we know.

What we do know is Paxwell was not so well planned.

I see no Paxwell on this one either.
 
Never heard of the term Paxwell? You must be new around here.

Dude who signed up in 2011 tells a dude who signed up in 2004 that he's new around here.

Go back to your alternate universe.
 
Dude who signed up in 2011 tells a dude who signed up in 2004 that he's new around here.

Go back to your alternate universe.

you can have a 20 years account here and still be newbie, what he was referring was "new" in the nvidia subforum anyone frequent in this sub-forum in the past months have heard the term Paxwell due the similarities of Maxwell and Pascal....
 
  • Like
Reactions: N4CR
like this
Wasn't there in issue in FAB for Nvidia where they already wanted to be on 16nm when the release of Maxwell came through? With that, I think in the back of my mind, that Maxwell was a stop gap to get something released (that was targeted for 16nm), but had to resort to 28nm to get to market on time. So I see it as Maxwell was WHAT Pascal should have been, and Pascal is what was Intended to be released, just with some more improvements since maxwell to bring in the new arch with the new release (such as deep learning etc..). That's why I think there a lot of similarities....
 
you can have a 20 years account here and still be newbie, what he was referring was "new" in the nvidia subforum anyone frequent in this sub-forum in the past months have heard the term Paxwell due the similarities of Maxwell and Pascal....

And I'm saying it's silly to project a meme onto a roadmap.
 
Wasn't there in issue in FAB for Nvidia where they already wanted to be on 16nm when the release of Maxwell came through? With that, I think in the back of my mind, that Maxwell was a stop gap to get something released (that was targeted for 16nm), but had to resort to 28nm to get to market on time. So I see it as Maxwell was WHAT Pascal should have been, and Pascal is what was Intended to be released, just with some more improvements since maxwell to bring in the new arch with the new release (such as deep learning etc..). That's why I think there a lot of similarities....

yes, originally Maxwell was intended to be built at 20nm, but the problems with the TSMC 20nm process first made maxwell to receive a delay, that caused the first split between Maxwell and Pascal and changed the roadmap to 28nm maxwell and 16nm Pascal, the second split was caused due the HBM/2 problems.. due that reason and the new development nvidia decided to go with G5X and save production issues or any other possible delay.
 
Wasn't there in issue in FAB for Nvidia where they already wanted to be on 16nm when the release of Maxwell came through? With that, I think in the back of my mind, that Maxwell was a stop gap to get something released (that was targeted for 16nm), but had to resort to 28nm to get to market on time. So I see it as Maxwell was WHAT Pascal should have been, and Pascal is what was Intended to be released, just with some more improvements since maxwell to bring in the new arch with the new release (such as deep learning etc..). That's why I think there a lot of similarities....

Maxwell was targeted for 20nm if I recall right. But that node turned out to be a dud. Nvidia also changed to a tick/tock model.
 
Back
Top