1080 Ti specs leak

I'm just not impressed. One of these isn't going to be powerful enough to drive 4K at 60 FPS with everything maxed out but two is going to be overkill for accomplishing that task. So I'm not really sure what the market is for this.

Guess I'll wait for the next architecture which *should* be powerful enough to drive 4K/60 with a single card.
 
I'm just not impressed. One of these isn't going to be powerful enough to drive 4K at 60 FPS with everything maxed out but two is going to be overkill for accomplishing that task. So I'm not really sure what the market is for this.

Guess I'll wait for the next architecture which *should* be powerful enough to drive 4K/60 with a single card.

this is likely going to be 10% slower than titan x . it's a tiny difference, theese complaints should be made against gp102 in general if anything

Also I hjadn't even considered the possibility of them using GDDR5 on an eventual ti, makes sense to distinguish it from titan x

Also judging by the OC headroom and the 1070 and RX480 which both use 8GT/s ram ICs I would say 9GT/s will be very easily doable, so the bandwidth difference between Titan X and this rumored 1080Ti will be rendered null once you consider SM numbers.

GTX 1080 20 SMs, 320GB/s stock = 16GB/s per SM

Pitan X 28 SMs, 480GB/s stock = 17.14 GB/s per SM

GTX 1080Ti 26 SMs, 384 GB/s stock = 14.75 GB/s per SM

GTX 1070 15 SMs, 256 GB/s stock = 17GB/s per SM

1080Ti with 9GT/s memory is 432GB/s = 16.61GB/s per SM
 
Last edited:
I think the current Titan is pretty much what a 1080TI will be. And the full GP102 will be released as a Titan Black or so with 24GB.
 
The OC3D article is sourcing WCCFTech. Just FYI.

upload_2016-9-15_9-6-7.png

I'm just not impressed. One of these isn't going to be powerful enough to drive 4K at 60 FPS with everything maxed out but two is going to be overkill for accomplishing that task. So I'm not really sure what the market is for this.

Guess I'll wait for the next architecture which *should* be powerful enough to drive 4K/60 with a single card.
A single Pitan X can already get you there in a lot of games.
 
Even if it is from WCCF it's a fairly sensible guess
 
From everything I read, it's not the number of shaders/cu's hurting Pascal. It's the bandwidth that's holding it back. (Per Anandtech's memory & GPU GHz scaling vs performance articles)
 
I don't understand how the market will take a card priced above the $700 mark that isn't "the best." Aren't people just going to buy a Pascal Titan X if they're already going this far?
 
would be a real shame to see reg gddr5 instead of gddr5x. Who knows though perhaps the higher bus width on the ti, is able to provide enough bandwidth with their memory compression tech that it doesn't matter which memory type they use.

I really wish AMD would release a high end card. Nvidia could use some competition in the high end, and I certainly don't want to pay 600$+ for a 1080 AKA a "midrange card" in terms of die size
 
Anyone willing to spend $900-ish on a GPU will probably be willing to spend $1200 on the Titan XP. So who exactly is going to buy this card?

I'd bet Nvidia doesn't even release a 1080Ti...doesn't seem to be any "gaps" in their current line up that a 1080Ti could help fill. Unless AMD releases Vega early and it is competitive with a 1080.
 
Anyone willing to spend $900-ish on a GPU will probably be willing to spend $1200 on the Titan XP. So who exactly is going to buy this card?

I'd bet Nvidia doesn't even release a 1080Ti...doesn't seem to be any "gaps" in their current line up that a 1080Ti could help fill. Unless AMD releases Vega early and it is competitive with a 1080.

The gap between a 1080 and Titan is 30-40% so right in the middle of that screams Ti. And if that leak serves right it is priced right too lol.
 
The gap between a 1080 and Titan is 30-40% so right in the middle of that screams Ti. And if that leak serves right it is priced right too lol.

There's a price gap there, but I don't know if there's any customers in that gap. Again, anyone with $900 to blow on a GPU wants the BEST, and if the Titan XP is 10-15% faster, they'll be willing to blow the extra $300 to get the BEST. The 980Ti was damn near 95% of a Titan XM, and far lower in price.

How many 1080Ti's would they sell at that price? Enough to offset the cost of development and marketing?
 
There's a price gap there, but I don't know if there's any customers in that gap. Again, anyone with $900 to blow on a GPU wants the BEST, and if the Titan XP is 10-15% faster, they'll be willing to blow the extra $300 to get the BEST. The 980Ti was damn near 95% of a Titan XM, and far lower in price.

How many 1080Ti's would they sell at that price? Enough to offset the cost of development and marketing?

I don't think it costs them much to release it. Nvidia are not working on slim margins lol, don't believe that its any hardship. This is strategic profiteering.
 
There's a price gap there, but I don't know if there's any customers in that gap. Again, anyone with $900 to blow on a GPU wants the BEST, and if the Titan XP is 10-15% faster, they'll be willing to blow the extra $300 to get the BEST. The 980Ti was damn near 95% of a Titan XM, and far lower in price.

How many 1080Ti's would they sell at that price? Enough to offset the cost of development and marketing?

There isnt anything to develop. The cards are already there. They are just sold currently as Tesla, Quadro and Titan.
 
From even before the "Titan P" was released, it looked like that current "Titan P" was designed to be the 1080Ti, but Nvidia changed labeling and pricing once AMD's Polaris was shown to be zero threat to the 1070 and 1080 lines.
 
Anyone willing to spend $900-ish on a GPU will probably be willing to spend $1200 on the Titan XP. So who exactly is going to buy this card?

I'd bet Nvidia doesn't even release a 1080Ti...doesn't seem to be any "gaps" in their current line up that a 1080Ti could help fill. Unless AMD releases Vega early and it is competitive with a 1080.

The scenario you described is basically a win win for Nvidia. It will come close enough to the Titan that $800-900 looks like a deal, so holdouts on the Titan will buy it, especially with custom cooler options available. People that still want the best of the best will get Titan still.
 
I don't think it costs them much to release it. Nvidia are not working on slim margins lol, don't believe that its any hardship. This is strategic profiteering.

It's not a question of their margins...They surely have enough cash on hand to put a card together and release it. But they would have to do the math and figure out if it's worth it. How much will it cost to market? Will it be Nvidia-only, or released to other OEMs to produce? How much will it eat into Titan XP sales? If the ROI isn't there, they won't release the card regardless of their margins.

There isnt anything to develop. The cards are already there. They are just sold currently as Tesla, Quadro and Titan.

They have to make and test a new PCB at minimum. Even if they don't make it themselves, they'll want to release a "reference" design which costs money to develop. Since they already have the 1080 and Titan XP to lean on, it certainly won't cost as much money, but there's still a cost. Plus marketing costs. The real question is are there enough potential sales out there to offset the cost? And how much would a 1080Ti cut into Titan XP sales?
 
From everything I read, it's not the number of shaders/cu's hurting Pascal. It's the bandwidth that's holding it back. (Per Anandtech's memory & GPU GHz scaling vs performance articles)
MNg8otN.png


memory barely makes a difference, also worth noting they can't fix the clock so this isn't the most accurate comparison.

+20% power and temp limit will affect the core clock stability
 
The scenario you described is basically a win win for Nvidia. It will come close enough to the Titan that $800-900 looks like a deal, so holdouts on the Titan will buy it, especially with custom cooler options available. People that still want the best of the best will get Titan still.

I get your point, but I don't know if the numbers would work out that way. There's plenty of people who view the 1080 as too expensive to begin with. Anything over that price is getting into niche territory pretty quickly, and it's not cost effective to have a big selection of hardware available in that niche.
 
It's not a question of their margins...They surely have enough cash on hand to put a card together and release it. But they would have to do the math and figure out if it's worth it. How much will it cost to market? Will it be Nvidia-only, or released to other OEMs to produce? How much will it eat into Titan XP sales? If the ROI isn't there, they won't release the card regardless of their margins.



They have to make and test a new PCB at minimum. Even if they don't make it themselves, they'll want to release a "reference" design which costs money to develop. Since they already have the 1080 and Titan XP to lean on, it certainly won't cost as much money, but there's still a cost. Plus marketing costs. The real question is are there enough potential sales out there to offset the cost? And how much would a 1080Ti cut into Titan XP sales?

They already have the chips, the pcbs, etc. It's like ppl are re-questioning the whole Ti line again. Haven't we been thru this with each succession of the Ti? Do yall need to wonder again if it's possible? Why do we need to become bean counters again and rewrite the book?
 
It's not a question of their margins...They surely have enough cash on hand to put a card together and release it. But they would have to do the math and figure out if it's worth it. How much will it cost to market? Will it be Nvidia-only, or released to other OEMs to produce? How much will it eat into Titan XP sales? If the ROI isn't there, they won't release the card regardless of their margins.



They have to make and test a new PCB at minimum. Even if they don't make it themselves, they'll want to release a "reference" design which costs money to develop. Since they already have the 1080 and Titan XP to lean on, it certainly won't cost as much money, but there's still a cost. Plus marketing costs. The real question is are there enough potential sales out there to offset the cost? And how much would a 1080Ti cut into Titan XP sales?

same PCB
 
Another thing to consider, what do you think the odds are that the 1080Ti chips are those that just couldn't quite make it for quality control purposes as a Titan P? The proposed specs show just a single quad unit of 256 shaders being disabled
 
By linear interpolation, this is within 1.5% of a Titan X at specified boost clocks (which is all we can go by as we have no idea how something speculated and unreleased would overclock, so we have to compare apples to apples.


It could be real, but I don't think they'll go there, not while they are still selling Titans. It would be a bone headed business move to undersell your own product that is still selling.

If these figures are real, I'm thinking its more likely it is just a product they have advanced in the pipeline and are ready to drop if - and only if - the competition comes out with something that challenges the 1080. We all know it is unlikely for AMD to be able to do this any time soon, but maybe that was less apparent when they started developing ti, and they just finished it as a jut in case.

Or these figures could just be made up.

I learned a long time ago to not pay any attention to pre-prelease rumors. If its not officially announced it doesn't exist and I will treat it as such until it is officially announced.
 
The WCCF leak is claiming GDDR5X which's a lot more plausible IMO; especially since OC3D doesn't seem to be linking to the forum post claiming otherwise.

Meh, I'd be more interested in a new rumor about when it's coming out ("Do I buy a 1080 now or wait for the Ti?") than twiddling around the edges on specs. From past history we know that regardless of the precise details, the answer to that one is: "So close to the equivalent Titan that it makes anyone who bought on look like a chump."
 
Meh, I'd be more interested in a new rumor about when it's coming out ("Do I buy a 1080 now or wait for the Ti?") than twiddling around the edges on specs. From past history we know that regardless of the precise details, the answer to that one is: "So close to the equivalent Titan that it makes anyone who bought on look like a chump."

That's based on the 980ti / Maxwell Titan comparison.

I'd consider that a bit of an anomaly, as it was a clear Nvidia overreaction to the upcoming Fury X parts which I guess they expected to be faster than they actually were. So they wanted to undercut AMD before they even came to market.

I could be wrong, but I doubt they'll be doing so again this time around, and if they do, we are likely looking at pricing close to the previous Titan's $1000 mark, not the fantastic deal the 980ti wound up being.
 
^^Truth, the 980ti had some competition. Today there is nothing on the horizon to drag this Ti's price down. And there won't be for a couple quarters, so I doubt there is a need for Nvidia to technically need one released.
 
They already have the chips, the pcbs, etc. It's like ppl are re-questioning the whole Ti line again. Haven't we been thru this with each succession of the Ti? Do yall need to wonder again if it's possible? Why do we need to become bean counters again and rewrite the book?

I'm not saying that can't make one, just speculating on if they SHOULD. The Nvidia bean counters have probably been working on this math behind the scenes for months now. If the ROI is positive, I'm sure we'll see a card.


Still will require some testing and certification if the leak is to be believed, since it's using GDDR5 vs 5x.

Another thing to consider, what do you think the odds are that the 1080Ti chips are those that just couldn't quite make it for quality control purposes as a Titan P? The proposed specs show just a single quad unit of 256 shaders being disabled

This right here would probably be the driving factor behind a release. If they have enough chips that work, but don't quite make the XP cut, a 1080Ti will be the only way for them to recover cash on them.
 
Yep. At this point it's strategic profiteering, displacing those that didn't jump on Titan with their moneys for the Ti. Can't get the Turbo? Here's a Carrera 4 for you.

Yep, I am sure Titan sales determine the exact release date of the 1080TI. But with that said, they want it out for the holiday sales if possible.
 
^^Truth, the 980ti had some competition. Today there is nothing on the horizon to drag this Ti's price down. And there won't be for a couple quarters, so I doubt there is a need for Nvidia to technically need one released.


Yep, the motivation to release a 1080ti would likely be less market driven and more driven by "we have all these GP102's that don't quite meet Titan X specs, it would be nice to make some money off of them".

Then they'd have the marketing and finance folks crunch the numbers on how much a 1080ti would hurt Titan X sales and how much they could get away with selling it for.

If it does launch any time soon (well before an AMD Vega) I'd expect it to be a $999 part, but I don't think it is even clear it will.
 
Oh to hell with it, I'm skipping this entire generation. Nvidia has gone full retard on their prices and AMD is sitting in the corner eating paste. If this is what the high end graphics card market is going to start looking like, I'm done with this hobby.
 
Back
Top