NVIDIA GTX 1080-successor By Late-July

[btarunr]

NVIDIA is reportedly giving finishing touches to its first serious GeForce-branded GPU based on a next-generation NVIDIA architecture (nobody knows which), for a late-July product announcement. This involves a limited reference-design "Founders Edition" product launch in July, followed by custom-design graphics card launches in August and September. This chip could be the second-largest client-segment implementation of said architecture succeeding the GP104, which powers the GTX 1080 and GTX 1070.

It's growing increasingly clear that the first product could be codenamed "Turing" after all, and that "Turing" may not be the codename of an architecture or a silicon, but rather an SKU (likely either named GTX 1180 or GTX 2080). As with all previous NVIDIA product-stack roll-outs since the GTX 680, NVIDIA will position the GTX 1080-successor as a high-end product initially, as it will be faster than the GTX 1080 Ti, but the product will later play second-fiddle to a GTX 1080 Ti-successor based on a bigger chip.



View at TechPowerUp Main Site

 

[Space Lynx]

Hmm. price of ethereum has been rising. I doubt any of us are able to get our hands on this. GDDR6 vram oc'd is probably going to be good for mining, maybe even more beneficial over an ASIC miner due to 70% power draw limit miners will put on the core. I am holding on to my 1080 ti until I have a 1180 ti in my hands. until then, never trust the markets, and miners can bugger off. i hate having to live like this now, blasted mongrels of the foul depths!

 

[dwade]

RIP AMD

 

[TheGuruStud]

Another paxwell, but with gddr6. I can't wait.

 

[dj-electric]

Another paxwell, but with gddr6. I can't wait.

Lolol shillvidia never stop rehashing lololol

 

[Xzibit]

So all they got out of their sources was

-A time frame that was already reported & process to production (Why did they have to speak anonymously about that)

The rest came from WCCFTech

 

[sutyi]

Another paxwell, but with gddr6. I can't wait.

I sort of agree with you there, but then again... Does it work on the gaming market? Yes, very much it does.

 

[ppn]

Better late than never

 

[Assimilator]

Lolol shillvidia never stop rehashing lololol

And yet, their three-and-a-half-year-old design (ancient in technology terms) still beats AMD's latest and greatest.

 

[dj-electric]

And yet, their three-and-a-half-year-old design (ancient in technology terms) still beats AMD's latest and greatest.

I know, that was a sarcastic WCCF-style comment in response to another.
The leap from Maxwell to Pascal yielded some impressive performance per watt numbers, and i can't wait to see what the next architecture can bring, considering Volta has already took half a gen step

 

[TheGuruStud]

I sort of agree with you there, but then again... Does it work on the gaming market? Yes, very much it does.

Only if you're a sucker. There's still no reason to move on from OCed 980ti. And now that they're even bigger D-bags and the available freesync monitor pool is getting good...no thanks. I will not be buying another.

 

[ppn]

12nm and 16nm is the same gen, but 7nm is another level. Will yield 2.5x density and 40% clock, so that 2.8 Ghz 1024 core GTX 1130 will deliver the same performance as GTX 1070. Cant wait for the 1170. 1180 will be overpriced initially and will take a big hit by the 1180 Ti.

 

[Vya Domus]

so that 2.8 Ghz 1024 core GTX 1130 will deliver the same performance as GTX 1070.

You wont see that anytime soon , 7nm isn't the miracle people think it is.

 

[Midland Dog]

Another paxwell, but with gddr6. I can't wait.

actually probably not, im pretty sure turing is volta without fp64, fp16 and tensor. Volta was highly different to both maxwell/pascal. It featured a proper hardware scheduler for the first time since fermi i believe. If tuned properly it will allow for full async support and should help out with d3d12

 

[TheGuruStud]

actually probably not, im pretty sure turing is volta without fp64, fp16 and tensor. Volta was highly different to both maxwell/pascal. It featured a proper hardware scheduler for the first time since fermi i believe. If tuned properly it will allow for full async support and should help out with d3d12

Downclocking the ram will prove one way or the other whenever they might appear (fall).

 

[happita]

12nm and 16nm is the same gen, but 7nm is another level. Will yield 2.5x density and 40% clock, so that 2.8 Ghz 1024 core GTX 1130 will deliver the same performance as GTX 1070.

That's a little bit of a stretch, even if it is 7nm. Performance over generations aren't that insanely increased with a node shrink. It's just enough to keep up with current gaming demands so they can milk us consumers as much as they can.

 

[GWComputers]

RIP AMD

Since has been 3 years that AMD is not competing with NVIDIA with GPUs and everybody knows this, i find your comment pretty lame and to no benefit for the topic?! I mean Rip amd is the best you can do for a topic that has nothing to do with comparison?! What a douche

 

[RH92]

actually probably not, im pretty sure turing is volta without fp64, fp16 and tensor.

Not so sure about the tensor core part since it's important for ray tracing .

 

[Midland Dog]

Not so sure about the tensor core part since it's important for ray tracing .

might have a few but expect it to be like fp64 on gp104/102/106/107 there but not really there like 1/32

 

[Vya Domus]

Since has been 3 years that AMD is not competing with NVIDIA with GPUs and everybody knows this, i find your comment pretty lame and to no benefit for the topic?! I mean Rip amd is the best you can do for a topic that has nothing to do with comparison?! What a douche

R.I.P AMD with their three fold increase in share price in the last few years.

 

[B-Real]

RIP AMD

As for the specs released in the TPUp data sheet, the 1080 should perform around the 1180Ti, which would be a big disappointment compared to Maxwell and Pascal, where the x70 cards beat the previous gen x80Ti cards. Of course it's speculation, but if it is the case, this would mean the exact opposite for AMD: a chance to catch the high-end GPU of NV. We'll see.

12nm and 16nm is the same gen, but 7nm is another level. Will yield 2.5x density and 40% clock, so that 2.8 Ghz 1024 core GTX 1130 will deliver the same performance as GTX 1070. Cant wait for the 1170. 1180 will be overpriced initially and will take a big hit by the 1180 Ti.

Whaaaat? 1130 same performance as the 1070? If these are you expectations, you will be greatly disappointed.

 

[M2B]

GTX 1180 Ti = 5120 CUDA Cores
GTX 1180 = 3580 CUDA Cores
GTX 1170 = 2685 CUDA Cores
GTX 1160 = 1790 CUDA Cores

 

[btarunr]

GTX 1180 Ti = 5120 CUDA Cores
GTX 1180 = 3580 CUDA Cores
GTX 1170 = 2685 CUDA Cores
GTX 1160 = 1790 CUDA Cores

My expectations are more along the lines of:

• GTX xx80 Ti: 5,120 CUDA cores
• GTX xx80: 3,072 CUDA cores
• GTX xx70: 2,304 CUDA cores
• GTX xx60: 1,536 CUDA cores

 

[bug]

It could be 1 (one) CUDA core for all I care. What matters is the overall performance.
Remember when AMD went all parallel, they discovered they needed async compute to be able to feed all those threads/cores. So number without context are as meaningless as ever.

 

[the54thvoid]

There's a review on TPU list today for a Titan Volta. It's held back at anything under 4k, but at 4k it generally improves over the 1080ti by 15-20%, higher in some cases. Overclocked it's an additional 15% faster. So, if TV will be the new 1180ti, it's about 30% faster than its predecessor, or, 50-60% faster than the 1080. Give or take.
I don't expect miracles with Nvidias next but I'm sure we'll see more of the same, generationally.