NVIDIA GeForce RTX 2080 Ti Founders Edition 11 GB

[Paganstomp]

No. Not a first day buyer here since I don't own or play any game that needs a HYBRID RAY TRACING card. This would be a complete waste of money. There are just too many things in life that need to be taking care of. But would be great for elitist snobs that tend to get themselves into inappropriate situations with 3D printed guns and under age Texas girls.

 

[dwade]

Low quality bait M8. This kinda post work better over on reddit.

There is no evidence that Navi will be PS5 exclusive. Sony may get early access if they are helping to finance it, but that does not mean it wont be made into PC GPUs.

AMD also didnt abandon the Pc for consoles. They released the polaris chips to great success, and released the half baked Vegas. They are not very competent at GPUs right now, but they are still there.

Also keep in mind AMD has way less revenue then either nvidia or intel, and is competing against both and doing SOMETHING. The fact the 480 was as close to nvidia's 1060as it was was a miracle. Now, with ryzen making serious dough and AMD finally clawing back some server sales, investment in RTG should allow a new arch that is at least somewhat competitive with turing.

https://www.forbes.com/sites/jasone...nys-playstation-5-vega-suffered/#31816a024fda
Vega suffered because AMD decided to reallocate their resources onto Navi for Sony's console. Raja left for a reason. AMD chose console gaming over PC gaming. End of story.

Back on topic. Solid review for a solid graphics card. If only there are more of them in stock since they're selling out like crazy.

 

[terroralpha]

...the cards are a total "winged unicorn." to those willing to buy the dream. Curious to see what the GTX 2070's bring to the table.

absolutely nothing. you will get 1080-like performance but the card will be way too slow for ray tracing so no one will ever be able to use it. to only upside will be that the 2070 will not choke to RX 580 speed anytime async is implemented. that's it.

 

[Caring1]

Raja left for a reason.

It's called corporate poaching, he has no loyalty, he prefers the almighty dollar, clearly.

AMD chose console gaming over PC gaming. End of story.

AMD chose to focus on the Processor side of their business for the short term, end of story, now back to the topic.

 

[Captain_Tom]

https://www.forbes.com/sites/jasone...nys-playstation-5-vega-suffered/#31816a024fda
Vega suffered because AMD decided to reallocate their resources onto Navi for Sony's console. Raja left for a reason. AMD chose console gaming over PC gaming. End of story.

Back on topic. Solid review for a solid graphics card. If only there are more of them in stock since they're selling out like crazy.

Vega was a well executed compute card, and if Raja couldn't see that - it's good he is gone. Bye bye.

AMD needs the PS5 WAY more than Vega needed that extra 20% performance. The revenue generated for AMD by the PS5 will allow them to truly fund a mega gaming architecture for the next decade, and that IS what they need.

I really don't get why that's so hard for people to understand. Furthermore, Nvidia is only as ahead as AMD lets them be ahead anyways. If AMD wanted to, they could launch an entire new line-up that's fairly competitive right now. However their money is better spend on EPYC.

 

[Assimilator]

Furthermore, Nvidia is only as ahead as AMD lets them be ahead anyways. If AMD wanted to, they could launch an entire new line-up that's fairly competitive right now.

Dafuq are you smoking?

 

[Captain_Tom]

Dafuq are you smoking?

What are you smoking? lol

"There's no way AMD can compete with the Titan!" Then 290X launches and roflstomps it with a smaller die. It's happened before, get a longer memory.

Things are definitely worse for Radeon right now - Turing is very impressive (Especially in my opinion), but that doesn't mean AMD couldn't launch an entire line-up by January that crushes the majority of Turing. It really wouldn't be very hard.

 

[R0H1T]

Dafuq are you smoking?

At 7nm TSMC, if AMD could get higher clocks out of Vega or refined Vega then yes it is possible for them to compete with Nvidia up to 2080 IMO. They won't touch 2080Ti anytime soon, unless they radically change their GPU uarch.

 

[W1zzard]

It really wouldn't be very hard.

The biggest issue for AMD is figuring out power efficiency as you can't just add "more stuff" like you do to raise performance

What's the biggest perf step AMD has ever made generation over generation? They need to double existing perf..
Once that's achieved they need to cut power of this new design into half, ie double perf/watt.. both at the same time

This is not like Ryzen where Intel made only small single digit percentage improvements generation over generation

 

[Assimilator]

What are you smoking? lol

"There's no way AMD can compete with the Titan!" Then 290X launches and roflstomps it with a smaller die. It's happened before, get a longer memory.

Things are definitely worse for Radeon right now - Turing is very impressive (Especially in my opinion), but that doesn't mean AMD couldn't launch an entire line-up by January that crushes the majority of Turing. It really wouldn't be very hard.

Either you're the world's most useless troll, or literally retarded. Where, exactly, is AMD going to procure this mythical faster-than-Turing architecture from, considering their current fastest GPU architecture is only half as performant as NVIDIA's, and they have made it very clear that there is no replacement for that arch arriving in the near future?

Unless you think that the shrink of Vega to 7nm will suddenly, magically make a poor architecture perform 200% better.... which it won't because that's not how it works. If it did, AMD would have been hyping it up like mofos and current Vega cards would already have been replaced.

But keep on with your delusions if that's what floats your boat, the rest of us will continue to live in the real world.

 

[btarunr]

No namecalling please.

 

[medi01]

What, exactly, is your point?

That greed was masked by "reasons".

NVIDIA isn't making life-saving medical supplies for everyone, they are producing a luxury product for a small market,

GPUs are not luxury products (not even high end) and GPU market is by no means "a small market", but it's again, has nothing to do with the discussion.

If you want prices to come down...

Why? I want them to go up.

This is not like Ryzen where Intel made only small single digit percentage improvements generation over generation

I think it's unrealistic to expect AMD's investment into high end GPU (way to risky).
Vega was a fluke (nVidai likely has a bunch and a half of own flukes which we have never seen, but unlike AMD it has enough money to fund R&D on multiple fronts), but 580 is a chip of roughly the same size as 1060 and it certainly doesn't consume "twice" the amount of power. (I also keep asking myself how many of the tested cards are cherry picked)

But given "full 7nm commitment", what is unrealistic about 1H 2019 Navi?

And regarding "Intel was doing nothing", uh, was it:

Steady progress through all these years, with Conroe jump looking that big only because P4 IPC was that low.

You should know better than me, that GPUs are inherently "more paralleled" and hence can still scale easier with die size, than CPUs.

 

[Assimilator]

Vega was a fluke

Wrong. See: Fury, Vega's predecessor. Not as bad as Vega, but GCN should have been retired then.

But given "full 7nm commitment", what is unrealistic about 1H 2019 Navi?

The fact that AMD has said it won't happen until 2H 2019 at best? Which means 2020 more likely, since their forecasts are always overly optimistic.

But even if they did manage to get 7nm Navi shipped in 1H 2019, GeForce RTX is available right now, which means it will benefit from two of the most important shopping events: Black Friday and Christmas of 2018. A GPU launching in 2019 misses both of those; a GPU launching in 2020 might as well not bother.

Vega's failure has caused AMD to give up on the high-end GPU market. Going forward, NVIDIA will be the only game in town in that bracket, unless Intel is somehow able to compete.

And regarding "Intel was doing nothing", uh, was it:

Steady progress through all these years, with Conroe jump looking that big only because P4 IPC was that low.

He didn't say "nothing". He said small improvements, which is absolutely true.

 

[medi01]

Wrong. See: Fury, Vega's predecessor.

Fury beat 980Ti at 4k, while Vega could barely take on 1080, so, no.

The fact that AMD has said it won't happen until 2H 2019 at best?

I'd appreciate a link and quote.

GeForce RTX is available right now, which means it will benefit from two of the most important shopping events...:

I'm not sure if it is stupidity caused by greed or some evil plan or both, but nVidai doesn't seem to consider dropping prices on older gen, 2xxx cards at given price are hard to justify and RTX stuff crippling 2080Ti to 1080p and poor fps is hardly exciting. And, uh, we also have that mining craze which I still can't figure if it's over or not and once it is over, we'd get market flooded with those cards, which is again, yay.

But even more importantly:
1) AMD doesn't have to "not miss" the mentioned shopping events
2) Exorbitant pricing set by Huang benefits AMD too
3) Mobile market is where they seem to have offerings.

He said small improvements, which is absolutely true.

Same improvements like before, while he implied Intel was slow, because no competition, so no, not rue.

Vega's failure has caused AMD to give up on the high-end GPU market. Going forward, NVIDIA will be the only game in town in that bracket, unless Intel is somehow able to compete.

After AMD recovers mid to mid-high end (which it could, having much more resources to spend on R&D than before), lucrative high end will naturally follow.

 

[Captain_Tom]

The biggest issue for AMD is figuring out power efficiency as you can't just add "more stuff" like you do to raise performance

What's the biggest perf step AMD has ever made generation over generation? They need to double existing perf..
Once that's achieved they need to cut power of this new design into half, ie double perf/watt.. both at the same time

This is not like Ryzen where Intel made only small single digit percentage improvements generation over generation

I said most of Nvidia's current line-up, and yes they really could launch 14nm chips right now with GDDR6 that easily compete with everything up to the 2080. Vega M in laptops has the same or better efficiency as Pascal, and heck we already know the XBX has a 384-bit Polaris card with extra compute units that would beat the 1070. But neither of these are on desktop because AMD is literally choosing to not care.

They launched a small midrange chip in 2016, and then a compute chip with half the ram in 2017 - that's it lol! Anyone acting like AMD has no options fails to see they haven't even tried.

Either you're the world's most useless troll, or literally retarded. Where, exactly, is AMD going to procure this mythical faster-than-Turing architecture from, considering their current fastest GPU architecture is only half as performant as NVIDIA's

LOL see my other comment - AMD has almost releasesd nothing for 2 years, and it's because that's all they need to do to keep 1/3rd of the "retarded" PC gaming market. They aren't even trying at all lol. But if you want me to tell you about this "Mythical" Turing-Killer - I need you to tell me which process AMD is using. There are multiple ways to skin a cat:

• GF 12nmFF
• GF 12nm-Planar
• TSMC 7nmFF

^ Choose. I could even conceive of a GF 14nmFF process that at least crushes Turing in Price/perf, but top performance would be hard to achieve using such an inferior node.

 

[efikkan]

At 7nm TSMC, if AMD could get higher clocks out of Vega or refined Vega then yes it is possible for them to compete with Nvidia up to 2080 IMO. They won't touch 2080Ti anytime soon, unless they radically change their GPU uarch.

7nm is not going to save AMD. Pascal already offers ~80% and Turing ~100% more performance per watt. Even the best node shrink will not offset that.

The reason why AMD's GPUs struggle is not lack of theoretical performance, it's lack of utilization. Throwing in more cores is not going to solve it either, doing so will only increase the gap to Nvidia. Look at what Nvidia have been doing; they don't just add more cores, they also keep squeezing more and more performance out of them, and that is ultimately also the key to energy efficiency.

Before Polaris arrived, people claimed it would put AMD ahead of Nvidia in efficiency, with the claimed superior 14 nm vs. 16 nm, but in reality it barely became on par with Maxwell on 28 nm. Now people are desperately clinging to the hope that 7 nm is going to save them, despite the efficiency gap being larger than ever.

 

[Captain_Tom]

7nm is not going to save AMD. Pascal already offers ~80% and Turing ~100% more performance per watt. Even the best node shrink will not offset that.

The reason why AMD's GPUs struggle is not lack of theoretical performance, it's lack of utilization. Throwing in more cores is not going to solve it either, doing so will only increase the gap to Nvidia. Look at what Nvidia have been doing; they don't just add more cores, they also keep squeezing more and more performance out of them, and that is ultimately also the key to energy efficiency.

Wake me up when Pascal can run make 15w GPU's for laptops that compete with the 1050 Ti...

 

[Durvelle27]

Definitely saw this coming.

Performance isn’t terrible but the price makes it’s very unappealing. Nvidia could have sold many more and had better support if it was priced better especially for those who have 980Ti-1080s. But at this price you’re better off buying a 1080Ti and Overclocking it.

 

[swirl09]

Ill answer your real(/hypothetical) question. If Im being honest, I would say Id have to see how I feel at the time, but I believe I would be leaning towards a no.

I know its probable that no one cares, but I want to change my answer. I am aware that most say whatever on the internet and its almost immediately discarded/forgotten, but I would like to think I stand over what I say - if only for myself.

With that said, the question was whether or not I would pay $1,500 for another 38% perf bump next time around. Quoted was my answer. Im changing this to a flat no. Having just paid only a little less than that for my still pending 2080ti, the thought of only paying a couple of hundred more to go another 38% (in the here and now*!) was tempting, but of course if I had to wait another gen AND pay more, it was a probably no.

The thing I did not consider, was that while a third or so more perf now (going to the 2080ti) is decent, part of my purchase was the due to the addition of the new cores. Even with them being (largely) unproven at this point, the prospect of what DLSS could bring was enough to tip the scales. So "just" getting another similar bump next time around at a higher cost would NOT do it for me.

 

[R0H1T]

7nm is not going to save AMD. Pascal already offers ~80% and Turing ~100% more performance per watt. Even the best node shrink will not offset that.

The reason why AMD's GPUs struggle is not lack of theoretical performance, it's lack of utilization. Throwing in more cores is not going to solve it either, doing so will only increase the gap to Nvidia. Look at what Nvidia have been doing; they don't just add more cores, they also keep squeezing more and more performance out of them, and that is ultimately also the key to energy efficiency.

Before Polaris arrived, people claimed it would put AMD ahead of Nvidia in efficiency, with the claimed superior 14 nm vs. 16 nm, but in reality it barely became on par with Maxwell on 28 nm. Now people are desperately clinging to the hope that 7 nm is going to save them, despite the efficiency gap being larger than ever.

I'm not saying anything about more (GPU) cores but higher clocks which can & probably will help Vega, again provided it clocks higher on 7nm without killing efficiency.

The assumption was that GF 14nm was gonna be superior to 16nm TSMC, which probably turned out to be untrue.

So you're saying 7nm TSMC will be worse than 12nm (SS?) or Vega is just bad regardless of what node it's on?

 

[efikkan]

I'm not saying anything about more (GPU) cores but higher clocks which can & probably will help Vega, again provided it clocks higher on 7nm without killing efficiency.

If we're lucky, we're looking at a choice between retaining the clocks and improving efficiency up to 50-60% or retaining the maximum TDP and pushing the clocks ~30%, or a mix of the two. One thing to remember is that hotter chips are harder to shrink, because watt per mm² also becomes a scaling threshold, leading them to either sacrifice density or clock speed. More efficient designs will benefit more from better nodes than inferior designs, that's why Nvidia increased their lead when going from 28 nm to 16 nm/14 nm.

When Turing have 100% more performance per watt and nearly 100% more raw performance, and all of this with just having 15% Flop/s than Vega 64, it should start to paint a picture of how far behind AMD is.

So you're saying 7nm TSMC will be worse than 12nm (SS?) or Vega is just bad regardless of what node it's on?

GCN is outdated and have fallen behind. 7 nm is not going to save them, and by the time AMD have 7 nm, Nvidia also will have access to it.

 

[medi01]

. Pascal already offers ~80%

Right:

Almost. Sort of kind of. But actually less, if we get back from reality distortion field of yours.

 

[newtekie1]

Right:

Almost. Sort of kind of. But actually less, if we get back from reality distortion field of yours.

In that chart, Vega64 vs GTX 1080 puts the 1080 85% better performance per watt.

 

[Diverge]

Anyone know the max power limit in afterburner/precision on the FE cards?

Evga released an updated bios for their cards, that takes them to 130% max power limit... so curious what the review sample FE cards have. If it is less, do you think nvidia will release a bios update? (do they ever release bios updates?)

https://forums.evga.com/tm.aspx?m=2858793

If they are limited to 115%, that seems odd, because nvidia specifically said that the 20 series has a robust power delivery, and more headroom, and previous gen FE cards were limited to 120%.

 

[Xzibit]

Anyone know the max power limit in afterburner/precision on the FE cards?

Evga released an updated bios for their cards, that takes them to 130% max power limit... so curious what the review sample FE cards have. If it is less, do you think nvidia will release a bios update? (do they ever release bios updates?)

https://forums.evga.com/tm.aspx?m=2858793

If they are limited to 115%, that seems odd, because nvidia specifically said that the 20 series has a robust power delivery, and more headroom, and previous gen FE cards were limited to 120%.

One of the youtubers forgot which one had one with 130%. Didnt do much different. As in peaked but then settled down. 204x-208x. Runs nice and cool.

It was jayztwocents @ 9:00

Nvidia already said they were voltage locked so going to need some hard workarounds