Nvidia GTX 1080Ti on its way to an outlet near you.

[P4-630]

I havent got $2000 to spend on this card

Bodyparts? Organs?

 

[Melvis]

Bodyparts? Organs?

Kidney it is!

 

[Frag_Maniac]

^^It's gonna be a while...

I know, but my point is, Vega is rumored to launch in the same time frame, 1st half of 2017.

It's sounding like Vega may eclipse the 1080 (at least on paper), but whether it can go toe to toe with the 1080 Ti, that's what really matters I think. It will all come down to price per performance as usual though.

 

[thesmokingman]

I know, but my point is, Vega is rumored to launch in the same time frame, 1st half of 2017.

It's sounding like Vega may eclipse the 1080 (at least on paper), but whether it can go toe to toe with the 1080 Ti, that's what really matters I think. It will all come down to price per performance as usual though.

IMO it should be competing vs the Ti or X. Because really, what's the point of HBM2, massive architectural improvements to only match the current/previous gen mid/upper tier card? And then Nvidia has their Pascal refresh or next gen in the works/wings. It had better be something to behold or... its gonna be the end.

 

[Kanan]

IMO it should be competing vs the Ti or X. Because really, what's the point of HBM2, massive architectural improvements to only match the current/previous gen mid/upper tier card? And then Nvidia has their Pascal refresh or next gen in the works/wings. It had better be something to behold or... its gonna be the end.

HBM2's reason is saving power while increasing bandwidth / performance. While Nvidia doesn't need it for their consumer cards because GDDR5X is sufficient, AMD needs it because their big GPUs consume too much power and are bandwidth hungry as fuck. Luckily they developed just the right technology to solve both problems for them - but this doesn't help their mid and low tier GPUs, as HBM2 is too expensive to be used there. It's just meant for enthusiast/high end products.

 

[Frag_Maniac]

It had better be something to behold or... its gonna be the end.

Agree, especially since Fury was kinda underwhelming.

 

[purecain]

Vega is going to compete with the 1080, it wont be anywhere near a 1080ti.
as far as the 1080ti vs the titan x goes.
quote

Titan X Pascal Minus 2 SMs Is Still A LOT of Horsepower
Inside Nvidia’s 12 billion 471mm² GP102 GPU that powers the GTX Titan X there’s a total of 3840 CUDA cores, grouped into 30 Streaming Multiprocessors arranged into sixGraphics Processing Clusters. Each SM houses 128 CUDA cores. The 1080 Ti having 3328 CUDA cores means we’re looking at four SMs fewer than the Titan X & eight fewer than what the full chip has . Inside every SM there are eight Texture Mapping Units and inside every GPC there are five SMs and a total of 16 Render Output Units. Cutting four SMs means the GTX 1080 Ti ends up with 208 TMUs as opposed to the GTX Titan X’s 224.

ive sold my Titan x and my radeon 290X so i'll be buying the titan I think.



HOLY SHIT I JUST DROPPED £1099 on a TITAN X pascal fuita edition. woooow

 

[64K]

My hope is that Vega 10 does come in somewhere close to a 1080 Ti in performance or some later refresh of it does. Otherwise we are looking at 2nd half of 2018 before we get 7nm Vega 20 and that's a long while after Volta comes which most likely means Voltas will be priced pretty high.

 

[EarthDog]

I hope so too, but certainly not expecting it. GTX 1080 performamce or a couple % better is my guess... and it will undercut it's price (until nvidia counters and lowers)

 

[Steevo]

What gets me is that the TXP uses a cut down GPU and this one will be cut down even further. Us PC gaming enthusiasts can't buy the full GPU at any price.

Yields..... this go around you can have great power and clocks with mediocre yields, or great yields with mediocre clocks and power.

 

[qubit]

Yields..... this go around you can have great power and clocks with mediocre yields, or great yields with mediocre clocks and power.

Yeah, I know. And the problem is likely to get worse as the chips get bigger, as in more transistors.

 

[Steevo]

Yeah, I know. And the problem is likely to get worse as the chips get bigger, as in more transistors.

Yep, perhaps soon though we will have another media to print onto with fewer defects, or shrinking node size will force changes in silicon production which results in fewer flaws, or even self assembling circuits.

 

[EarthDog]

What is Titan X Pascal cut down from? It does have its own core (gp102) compared with gp100 from the Tesla P100, however, the gp102 has MORE (but different - fp64) CUDA cores than P100 (3840 vs 3584 of P100). Along with the P100 having HBM2 and a shed load of it, It also has double precision abilities while X Pascal does not, right?

So I am curious as to what the Titan XP is cut down from....I think it is it's own silicon and the 1080ti will be cut down from it/GP102.

 

[qubit]

Yep, perhaps soon though we will have another media to print onto with fewer defects, or shrinking node size will force changes in silicon production which results in fewer flaws, or even self assembling circuits.

Yeah, here's hoping. The end of Moore's law might mean the breakthrough of a new technology that's been waiting in the wings to give us great leaps in performance, but I wouldn't hold my breath for it. These things will happen on their own schedule, not when we want them.

What is Titan X Pascal cut down from? It does have its own core (gp102) compared with gp100 from the Tesla P100, however, the gp102 has MORE (but different - fp64) CUDA cores than P100 (3840 vs 3584 of P100). Along with the P100 having HBM2 and a shed load of it, It also has double precision abilities while X Pascal does not, right?

So I am curious as to what the Titan XP is cut down from....I think it is it's own silicon and the 1080ti will be cut down from it/GP102.

If you check out TPU's review, you can see that it's not the full GPU. I'm not actually sure if it's a GP100 core cut down or a slightly different design, but regardless some of it is disabled.

https://www.techpowerup.com/reviews/NVIDIA/Titan_X_Pascal/2.html

 

[EarthDog]

Thanks for the link... funny, that was the article I was looking at and missed the picture... haha!

I can't be a GP100 core... it has LESS CUDA cores than GP102.

EDIT: Reading would help...!!

The GP102 is essentially GP100 minus FP64 CUDA cores, featuring only 3,840 FP32 ones, and 4096-bit HBM2 memory interface making way for 384-bit GDDR5X. As mentioned earlier, the GP102 is a 1.5X upscale of the GP104 silicon, which powers the GeForce GTX 1080.

So, in essence, that picture is GP100 with the FP64 CUDA cores eliminated making it a Titan. So, it is cut down, however, its only the FP64 cores for double precision/SM it seems???

 

[purecain]

The availability on the titan is touch and go. I paid for the order yesterday and it has still not shipped with customer services saying they have no idea when it will ship.
A very poor buying experience indeed. hopefully it will be here by next week. but just like customer services says, there is just no way of knowing. ^^

I got my card and its amazing.

 

[Kanan]

Thanks for the link... funny, that was the article I was looking at and missed the picture... haha!

I can't be a GP100 core... it has LESS CUDA cores than GP102.

EDIT: Reading would help...!!



So, in essence, that picture is GP100 with the FP64 CUDA cores eliminated making it a Titan. So, it is cut down, however, its only the FP64 cores for double precision/SM it seems???

The full GPU is only found on Quadro products, GM102 that is (basically Titan X-P is expensive and good working garbage). Tesla has the GM100 620mm² monster chip with both DP and SP shaders.

 

[The N]

the question is how much time they will give to 1080Ti user untill Pascal refresh launches

 

[the54thvoid]

the question is how much time they will give to 1080Ti user untill Pascal refresh launches

If Vega beats the 1080 then they may release 1080ti prior to refresh or if they feel Vega isn't enough then they may release the 1080ti as a refresh. Then again, it's Nvidia so you never can tell what they'll release.

 

[RejZoR]

They'll attack AMD's Vega with it (if needed, depending how Vega will even stack against vanilla GTX 1080).

 

[Kanan]

Users will win anyway, competition drives the market and prices. That said I expect Vega to be faster than 1080 but not faster than GP102.