Move Over GTX 1080, There’s A New Titan X In Town

[alidan]

I'm very surprised that the new Titan X has been revealed quite so early especially with how high demand is for 1080/1070/1060.

I'm not surprised: release the Titan before 1080 demand dies down and you can upsell Titans to big-walletted enthusiasts who get tired of waiting for 1080 availability. If enough chip production gets diverted to Titan X/1080Ti to meed demand to reduce 1080 availability, it may cause inflated retail prices to hold much longer.

Does that even benefit Nvidia monetarily? I don't know how the relationship with GPU partners is structured. One would presume that it's like most industries, and the manufacturers buy the chips at a negotiated rate, then resell to distributors and so on down the chain. In this scenario, Nvidia would make the same amount per chip delivered regardless of retail pricing. Now, if it's somehow royalty-based, and their cut rises with sale price of cards, manipulating retailer prices would make sense. The other scenario I could see is if GP104 supply is tight and/or yields still low, and pushing some GP102s out the door keeps the cash coming in.

An other other possibility is that AMD has an upper-midrange or even high-end offering coming sooner than people expect, and Nvidia is trying to keep the spotlight on themselves. It's not all that likely, but would fit with their strategy of positioning themselves as the premium option. On the subject of launch strategies (and diverging a bit), that could be why the RX 480 seems like it wasn't quite done; it may not have been, but AMD needed to launch it ahead of the 1060 to keep Nvidia from (once again) stealing all the thunder.

Most likely that whats going on. Nvidia usually has its finger on AMD's pulse. I hope that is the case I would like to see a bit more of a battle to combat these higher prices.

Honestly, Its like nvidia is jumping at ghosts, Instead of pushing lower end, smaller cards first to test the process, they push the big ones and get crap yields.

honestly, i'm assuming that amd is doubling the size of, at the very least, the 480 for the 490, possibly more then that if they also put out another fury line. amd loses out in dx11, but makes gets a fairly massive boost through rx12 and vulcan.

To me it looks like nvidia was trying to get their high margin cards out the door as fast as they could so they are nowhere near amds vega lineup, and the 480 forced nvidia to push the 1060 or lose the mid range buyers. It seems like stock 480 nitro outpaces a 1060, how easy these are to get to the 1420 mhz mark we will see, but that outpaces a 1060 2100 in most things, not to mention doom, where even stock out paces a heavy oc 1060, and this is what i anticipate dx12 games to come to do, so nvidia wants to get their better on 11 cards out fast to wait this gen out for volta where they will either have async capabilities, or they will have to have just that much bigger a chip all around.

really wish amd would get their crap together with dx11 though, i have a phenom 955 and i'm waiting on zen, would be nice to get some free performance.

 

[mike_233]

More cores would entertain from me, justification for a tag of 1200, if it entailed HBM, what was two years a 'promised' in 'Pascal'.
I'm thinking we won't see HBM till Volta where Cores and HBM meld together giving some serious increases in speed.
But then, Volta 'Titan' will probably be tagged at 1500$ also:\ Along with entirely new architecture to everything but DDR4 memory for mainboard Ram and power supplies.
time to start saving 5K for that eventuality.

 

[phytoplankton]

I wish Nvidia would come up with a better naming convention than Titan black, X, new X? It makes it pretty much impossible to search and find detailed info on a specific model. I have 2 of the first titan models and google is NOT my friend when I'm trying to look for projects to reuse them in.

Nvidia did it so the old cards hold its value. Even though the OG Titan is a little faster than a GTX 780, they still sell for $400+. They are completely worthless now. I was just playing with someone online yesterday and they were telling me how their Titan is the fastest and the best on the market. They weren't talking about the X either lol, just the OG one. I didn't have the heart to break it to them that my 1070 was more than twice as fast and cheaper than he paid for a used Titan.

GTX Titan = Something betweena 780 and 780TI
GTX Titan Black = IS the GXT 780 TI with 6GB of vram
GTX Titan X = GTX 980 Ti with 12gb of vram. (nothing more)

That's the reason I'm not planning on selling my old Titans. I'll just reuse them in a secondary PC or in some other project. I was looking for some ideas online but googling Titan and Titan SLI returns mostly pages regarding the later versions and excluding 'x', 'z' or 'black' in the search doesn't help much

 

[bit_user]

I'm thinking we won't see HBM till Volta where Cores and HBM meld together

Where did you read that?

That's the reason I'm not planning on selling my old Titans. I'll just reuse them in a secondary PC or in some other project.

Perhaps this might get you started:

http://www.nvidia.com/object/gpu-applications-domain.html

FP64-intensive applications should include computational fluid dynamics, structural mechanics, financial modelling, and climate simulations, for starters.

 

[TechyInAZ]

3440x1440 is the go to gaming pinnacle right now, not 4k. 4k will get skipped and 5040x2160 will be the standard as seen by testing from asus and acer. the decline of 16:9 has begun. long live 21:9

Yep, then in about a year or two we'll have 8k screens go mainstream. There not that far off, we already have plenty of 8k screen demos across the world.

 

[Samer1970]

3440x1440 is the go to gaming pinnacle right now, not 4k. 4k will get skipped and 5040x2160 will be the standard as seen by testing from asus and acer. the decline of 16:9 has begun. long live 21:9

meh .. I prefer 16:10 no ty .. you guys are so crazy about wide screens and you forgot Vertical Sight all together !

up and down is as important as left and right !

keep it 16:9 or even better 16:10 and make it large to span left and right, a 4K 40 inch !!! and zoom the game out a little so you can view more to the left and right. this is alot better than 21:9

 

[jasonelmore]

GPU core count for Nvidia has not scaled perfectly in the past. On top of that, the Titan X is clocked 20% slower, so 40% more cores but 20% slower, and your looking at around 20% gains over a 1080.. and thats if cores scale perfectly (which they dont)

The 1080Ti is looking to be a disappointing product without the HBM2. Nvidia is gonna be FUBAR when AMD's 2nd Gen HBM2 card comes out.

 

[jdwohlever]

Meh... I made the mistake of buying a first gen Titan only to get my butt handed to me 6 months later with the 980Ti for HALF the cost and half the features.
Titans are for people doing Light 3D and video editing I guess?
Even 3D is a stretch since the OpenGL implementation isn't very strong and when the Titan first came out there was no 3D apps that took advantage of CUDA.
Now, 3 years later there are some benefits to having a Titan for 3D, but most pro 3D people are going to go with a dedicated OpenGL card anyway for $3000 or more.
Ti's and lower are for gaming.

I don't really see the point of Titan at the end of the day.
Feature wise it's not as high end for gaming for cards half it's price.
For graphics work and workstation purposes it's actually pretty medium to low end.
I think the Titan is geared at two people:
1) Suckers like me who thought the higher VRAM and CUDA cores would matter for gaming (It didn't).
2) Workstation people who develop games/content specially for NVIDIA feature sets.
Maybe if you are developing a game that use Gameworks or CUDA features would having a Titan be useful?

 

[Samer1970]

Meh... I made the mistake of buying a first gen Titan only to get my butt handed to me 6 months later with the 980Ti for HALF the cost and half the features.
Titans are for people doing Light 3D and video editing I guess?
Even 3D is a stretch since the OpenGL implementation isn't very strong and when the Titan first came out there was no 3D apps that took advantage of CUDA.
Now, 3 years later there are some benefits to having a Titan for 3D, but most pro 3D people are going to go with a dedicated OpenGL card anyway for $3000 or more.
Ti's and lower are for gaming.

I don't really see the point of Titan at the end of the day.
Feature wise it's not as high end for gaming for cards half it's price.
For graphics work and workstation purposes it's actually pretty medium to low end.
I think the Titan is geared at two people:
1) Suckers like me who thought the higher VRAM and CUDA cores would matter for gaming (It didn't).
2) Workstation people who develop games/content specially for NVIDIA feature sets.
Maybe if you are developing a game that use Gameworks or CUDA features would having a Titan be useful?

3- to take your money while hiding the "Ti" few months 😛

 

[falchard]

The more nVidia 'releases' the more I feel this generation is a pass for nVidia as they work on a new architecture. They are just trying to shore up market share by releasing small quantities of cards that promise big things but are not available. It is somewhat effective at stopping potential AMD sales as they release a chip as soon as its available instead of building up stock.

 

[hatib]

if it is 50% faster then titan x then it is 25% then gtx 1080 is which is a big gain i expect this card to run 5k@60fps ultra settings without AA as for the pricing is should be 25% more expensive then gtx 1080 rather then 50% when you are getting 25% more performance then gtx 1080.Gtx 1080 msrp 25% is $150 so it should be $750 and if u take founders edition price then 25% is $175 so the price should $875 the max price for this gpu should $900

 

[Gillerer]

...
I was looking for some ideas online but googling Titan and Titan SLI returns mostly pages regarding the later versions and excluding 'x', 'z' or 'black' in the search doesn't help much

Maybe try googling:

gtx titan -"titan x" -"titan z" -"titan black"

Optinally add "SLI" in there, maybe "Nvidia"? And try with "Nvidia" but without the "GTX". I'd advise always using either "Nvidia" or "GTX" together with "Titan" to avoid getting pages referring to the monsters in Roman mythology.

Even then, one problem that ofter arises is that even if an article is originally from around the time the original Titan came out, various websites have links at the bottom to newer articles that it thinks are for the same subject matter. Such article titles may well contain "Titan X" or "Titan Black" etc., so the pages would not be found using the search above.

Another option if you want to see pages created around when the original Titan was new, is to use Google's time restriction. Since GTX Titan Black came out 18th Feb 2014 and GTX Titan Z on 25th March 2014, restrict the pages to time periods up until, say, 31st Jan 2014:

Google search
-> Search tools
-> Any time
-> Custom range
-> enter the "To" date in your Google locale's time format.

 

[bit_user]

... titan x then it is 25%
...
if u take founders edition price then 25% is $175 ...

I think you meant "than".

Cool theory on performance-based pricing, but I think they're charging the amount they judge will maximize profits, based on supply, demand, and the (lack of) competition.

In the end, I think the only ones who really base decisions on performance per dollar are people doing GPU-compute. And they also look at lifetime operating costs, so power efficiency tends to overshadow initial purchase price.

 

[Uniblab]

Just want to see if one Titan is better than 2x1080. If so then it will be a monster card. If not.....welll, people do drive lambos to work.

 

[Gillerer]

Depends on the game. If a game can't use SLI then the new Titan will be vastly better. The best a GTX 1080 SLI can hope for is to tie or just edge past in SLI-optimized titles. Because this GTX 1080 SLI doesn't really make sense in the same price range as the Titan.

 

[Aspiring techie]

One thing I'd like to see is an fps per clock per core comparison between AMD and nVidia. I want to see if nVidia hit those high clocks by sacrificing some efficiency per clock. At the very least, it would be interesting to see.

 

[turkey3_scratch]

One thing I'd like to see is an fps per clock per core comparison between AMD and nVidia. I want to see if nVidia hit those high clocks by sacrificing some efficiency per clock. At the very least, it would be interesting to see.

That would be very interesting to see!

 

[Noah_22]

Good thing we finally beat the Xbox One S. Phew!

 

[Samer1970]

Good thing we finally beat the Xbox One S. Phew!

whos "we" ???

 

[Jason_154]

Well... doesn't that mean Huang won the bet? He didn't think a GPU could produce 10 terafops with a single gpu. Kellher didn't. He produced 11.

 

[Aspiring techie]

One thing I'd like to see is an fps per clock per core comparison between AMD and nVidia. I want to see if nVidia hit those high clocks by sacrificing some efficiency per clock. At the very least, it would be interesting to see.

That would be very interesting to see!

The thing that caught my eye was that the RX 480 which is clocked under 1300 Mhz is only somewhat slower than the GTX 1060 which is clocked at over 1700 Mhz.

 

[InvalidError]

The thing that caught my eye was that the RX 480 clocked under 1300 Mhz is only somewhat slower than the GTX 1060 clocked at over 1700 Mhz.

30% is a pretty significant gap when you consider that very few CPUs and GPUs can achieve that sort of overclock through conventional means and the RX480 does not appear to be a particularly good overclocker to begin with.

 

[TechyInAZ]

The thing that caught my eye was that the RX 480 clocked under 1300 Mhz is only somewhat slower than the GTX 1060 clocked at over 1700 Mhz.

30% is a pretty significant gap when you consider that very few CPUs and GPUs can achieve that sort of overclock through conventional means and the RX480 does not appear to be a particularly good overclocker to begin with.

I believe the RX 480s reference PCB is at fault for bad overclocking. I think the RX 480 should get a lot longer legs once the 8 pin aftermarket cards come out.

 

[InvalidError]

I believe the RX 480s reference PCB is at fault for bad overclocking. I think the RX 480 should get a lot longer legs once the 8 pin aftermarket cards come out.

Time will tell. For now, reaching 1.7GHz on an RX480 requires a volt mod and liquid nitrogen.

 

[Gillerer]

GPU per-clock-per-"core" comparisons don't mean anything between vastly different architectures. One might be built to use twice the number of "cores" in order to arrive at similar performance.

Since GPU workloads generally scale pretty well to large number of cores, this doesn't affect the performance profile in the same way as with CPUs where depending on the workload you want high-power cores (but don't need many of them), or want maximum number of cores (with lesser focus on single core performance).