• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Pascal Titan-X Launch

AMD is not nvidia and look how that 4Gb move worked out for them, plus it was a limitation of HBM1.

So your saying that 18 months on every one expected the next titan to have the same amount of VRAM as the Titan X?

280 - 1gb
480 - 1.5gb
580 - 1.5gb (5 series was a refresh of the 480 series)
680 - 2gb

780 - 3gb
Titan - 6gb

980ti - 6gb
Titan - 12gb

As for GDDR5X being so special, the 1080 got double vram over the 980 as did the 1070 over the 970. So why is the this round of titan not got the jump?

Erm, because it's not really a measure of performance and is pointless?

Cue post about playing games with three/four cards with MSAA cranked up.

Cue Reality
 
There willl most likely be a 1080Ti... It is very obvious because they do it every time...

The reason for this is :

They do not get full yeilds on the Titan X...

They then save up these cards and release them later as the xx80 Ti with less CUDA cores.

Unless the titan X is already cut down and does not have the full cuda cores...

If when they release the reviews we see that the Titan X already has some disabled shader units, then it is possible there will not be 1080ti... but I would say that is very unlikely.
 
Last edited:
why do people emphasize Volta so much ? Pascal was always slated to be the next big one, is volta a new architecture or something ? can it take better advantage of the new process ?

Could they be holding back on HBM2 so that for Volta they can just add HBM2
and suddenly get 25% more performance with no effort ?

Wouldn't it be interesting if Vega has HBM2 but 1080ti doesn't ?


It was either Jayztwocents or Awesomesauce latest video where they said that Volta is expected in about 8-9 months time.
 
GP100 has same amount of cores, though.

GP102 is a reworked GP100 card dedicated for gaming. GP100 was not used because it is not suited for gaming.

Or are you saying that Nvidia have an entirely different GPU in development that we haven't heard about yet? Cuz I dont think that's the case. And it would totally water down the Titan name if they did.

From what it looks like is the GP102 is on par with the telsa version ( same cores)with all the stuff not needed for gaming stripped out which would make the chip smaller than the telsa version I think some where around 454mm, Now nvidia can still make a large chip as they can go to 600mm which makes room for the gaming GP100 which has the same number of transistors at 15.3 Billion but all for gaming. so its just a case of utilizing the space effectively for the task. In terms of watering down a name, all anyone should ever care about is how much performance they get for there money, and if it makes them happy then so be it.
 
why do people emphasize Volta so much ? Pascal was always slated to be the next big one, is volta a new architecture or something ? can it take better advantage of the new process ?

Could they be holding back on HBM2 so that for Volta they can just add HBM2
and suddenly get 25% more performance with no effort ?

Wouldn't it be interesting if Vega has HBM2 but 1080ti doesn't ?

there are some things we can read about it. not sure how accurate they are, of course.

http://growthfinancial.net/2016/07/...launch-first-volta-processors-in-may-of-2017/
 
Erm, because it's not really a measure of performance and is pointless?

Cue post about playing games with three/four cards with MSAA cranked up.

Cue Reality


Titan cards have always had pointless amounts of vram, its one of the reasons its priced as high as it is.

This is also not the point I was trying to convey, seanspeed said that its the same quantity of vram but GDDR5X infering that the GDDR5X was the reason it wasnt more than 12GB and made up for the lack of high quantity of VRAM. I was mearly pointing out that the 1080/1070 had doubled the VRAM of the previous gen and was also GDDR5X.
 
Last edited:
Hmm well I was debating going 1080 SLI but haven't brought yet because Parvum keep delaying case. This has me interested obviously won't be a powerful as 1080 SLI when it's properly supported but we all know how flaky SLI is still.

Would be Watercooling this or Watercooling the 1080 SLI

What do people think New Watercooled Titan X or 1080 SLI.

You have been very sensible holding off on ordering the 1080s before the case arrives. I was starting to worry it was just me waiting, will be 3 months next week since I ordered a Parvum L1 and still not received it. Had a pair of 1080s sat waiting to go in for a while now. Looks like the Titan X will release before I even get the 1080s in a case which is a bit of a slap in the face.

Still could be worse, I could still be waiting for the ASUS Strix 1080 OC to arrive when the Titan releases if I hadn't swapped already.
 
AMD is not nvidia
So I answered your question yet I'm not surprised you found a way to try and dismiss it anyways....great.

and look how that 4Gb move worked out for them, plus it was a limitation of HBM1.
It worked out fine. Fury was a disappointment for reasons other than its memory.

So your saying that 18 months on every one expected the next titan to have the same amount of VRAM as the Titan X?
I'm saying the switch to GDDR5X and improving memory management algorithms makes it an equally big jump. Here's some numbers to assure you leaps are being made:

Kepler Titan
6GB GDDR5
6Ghz
288GB/s

Maxwell Titan
12GB GDDR5
7Ghz
336GB/s

Pascal Titan
12GB GDDR5X
10Ghz
488GB/s

The switch to GDDR5X is extremely relevant in not needing to throw more modules on there.

why do people emphasize Volta so much ? Pascal was always slated to be the next big one, is volta a new architecture or something ?
Other way around. Volta is the big new architecture. Pascal was not originally in plans, but was thrown in as a stop-gap architecture to have something for when the 14/16nm process was ready to go.
 
Titan cards have always had pointless amounts of vram, its one of the reasons its priced as high as it is.

you are not the first one who keeps saying pointless.

If you don't need it then fair enough. Gibbo was saying that Titans were being snapped up by corporations and I am pretty sure it's not to feed their employee's gaming habits.

My point, it's not pointless to others and there are other uses for those other than gaming.
 
why do people emphasize Volta so much ? Pascal was always slated to be the next big one, is volta a new architecture or something ? can it take better advantage of the new process ?

Could they be holding back on HBM2 so that for Volta they can just add HBM2
and suddenly get 25% more performance with no effort ?

Wouldn't it be interesting if Vega has HBM2 but 1080ti doesn't ?

Pascal was slotted in as a short term solution, and is basically Maxwell on a smaller process (16nm instead of 28nm)

Volta will be a much bigger and more significant architectural change.
 
From what it looks like is the GP102 is on par with the telsa version ( same cores)with all the stuff not needed for gaming stripped out which would make the chip smaller than the telsa version I think some where around 454mm, Now nvidia can still make a large chip as they can go to 600mm which makes room for the gaming GP100 which has the same number of transistors at 15.3 Billion but all for gaming. so its just a case of utilizing the space effectively for the task.
Right, you're suggesting they have a completely new GPU in the works that nobody has gotten word of......and who wouldn't have a place in the naming scheme as GP100, 102, 104 and 106 are all taken...

In terms of watering down a name, all anyone should ever care about is how much performance they get for there money, and if it makes them happy then so be it.
I'm talking about from Nvidia's side. The Titan name is meant to be their flagship product. If they come out with a newer, bigger GPU later on than the Titan branding means nothing anymore. It's just another model in the lineup. Defeats the whole purpose of a 'flagship' product in terms of how a company uses it to boost brand perception.
 
Wow, this was a surprise. I was not expecting this so early. If it's up to 60% faster than an original Titan X, then according to my calculations, if correct, it's only up to 20-25% faster than a stock 1080 which isn't much for the price. I used this techpowerup review to get this https://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_1080/.

As others have said, this card may be what would traditionally be the 1080Ti and we'll see a true Titan card later with HBM2 for even more money called the TitanXX or whatever as Nvidia have no competition in this segment. This would be an inspired move by Nvidia, however, not good for the consumer or their wallet.
 
Everybody knows the script:
1.)Release the Titan X - people get excited and pay mega-monies for it since it is the "fastest card in the world"
2.)A few months later the GTX1080TI is launched which is faster and much cheaper

The thing with 2 is it isn't actually faster it simply does not have Nvidia holding it back by stopping the creation of overclocked non reference versions by there board partners.


The only time I remember that not being the case was when OCUK & INNO3D did cherrypicked Titan Black chips with the Hercules cooler at the end of it's life. But that was when the ti did have the same cuda cores and only lacked the additional 3gb's of vram.
 
Back
Top Bottom