Nvidia ,Rtx2080ti,2080,2070, information thread. Reviews and prices September 14.

[alcoholbob]

They are doing 2080 vs 1080. No Ti is involved.
And No, 2080Ti is going to be at least 50% faster than 1080Ti.

50% faster than 1080 Ti in Sniper Elite 4 maybe...where Titan V is already 65% faster lol

 

[cmdrdredd]

At least this gives me some hope it will turn out alright, just have to wait for the prices to drop to a more normal price

Which probably wont happen unless the cards are terrible at mining. My gut tells me that since people are buying 1080ti cards at $1000+ a pop they decided to put the new ti cards at around that for the MSRP since they figure it'll sell out anyway. It's pretty ridiculous.

 

[dlerious]

So just like Vega launch, I will be taking screenshots and showing them to the know it all crowd who just like to baselessly crap on any Turing news for no good reason except AdoredTV and AMD friendlies said so.

First you posted this link https://videocardz.com/77696/exclusive-nvidia-geforce-rtx-2080-ti-editors-day-leaks which says
"So you are all asking, why there are no performance leaks. The reason is simple: the very few who already have the cards simply can’t test them — there are no drivers. The worst part is that samples won’t ship to the rest of the reviewers until early September (!). Don’t expect performance leaks by that date, even if there are any, they may not represent the full potential of new cards."

Now I'm supposed to believe HardwareCanucks because they've been "briefed by Nvidia". How could they test the performance if there are no drivers?

 

[sze5003]

Mining craze has pretty much declined I think by now. But I do think when they saw people buying cards at the ridiculous prices they were inflated to due to miners, they of course saw a chance to make money off of it. So here we are and they are pushing the new architecture as the reason when in reality I don't think it will be all that performant in the beginning. I'm all for it but I don't think we are fully ready for it, it seems pretty expensive in games and if not done correctly will make the game suffer.

Which probably wont happen unless the cards are terrible at mining. My gut tells me that since people are buying 1080ti cards at $1000+ a pop they decided to put the new ti cards at around that for the MSRP since they figure it'll sell out anyway. It's pretty ridiculous.

 

[moonbogg]

There's no upgrade here for 1080Ti people. For $100 more than we paid for the 1080Ti, we get the RTX 2080 which trades blows with it, lol. Well, for me its really simple. If I am going to buy a new card, someone has to offer something better for the same money. That's how technology works and that's what I expect. So, that either happens or it just doesn't.

Nvidia takes a page from EA's playbook and tells us, "Accept it, or don't buy the card". Well alrighty then! That was easy!

 

[Malogeek]

There's no upgrade if you don't care about raytracing enhancing modern games. If you're interested in it from what we've seen then the reviews should give us a clear indication of the visual gains we are getting for what performance penalty and whether you're willing to pay that price. Making a decision on raytracing from what we've seen so far is like buying a Vega based on AMDs "feels fast" campaign.

 

[alcoholbob]

There's a new slide claiming shading performance gains vs Pascal. However it doesn't seem to overlap exactly with the GTX 2080 vs 1080 graphs. Since this graph doesn't specify which cards are being compared, this could be an apples to oranges comparison.

 

[zinfamous]

They are being stupid and dense this time, I don't know why.probably because of RTX.

They have increased efficiency by 50% core to core, and have the 2080 being 10% faster than 1080Ti. So 2080Ti being only 10% faster than 2080 is nonsense. The difference between them in raw specs is large.

There's no data points in those slides that you keep posting, and there is no key as to what is being compared as 1.5x or 2x of...what?

Those are absolutely dumb charts. They don't say anything until we see actual data. "Oh Look! the performance of a 2080 is 2x of a 1070!" ...would be as equally true of anything else that this chart is purportedly claiming.

 

[zinfamous]

LOL AdoredTV is the bible now? he said 50% in the previous leak, now he says 20%. he can't make up his mind.

It IS 50% because this is what NV is declaring, he can't possibly know the product more than NVIDIA can he?

what is 50% faster? A 2080 over a 1070? I would say that much is likely true. One would hope it is much, much faster though. 50% greater efficiency, core to core, isn't 50% greater performance. Now you're using RTG marketing.

 

[neblogai]

There's a new slide claiming shading performance gains vs Pascal. However it doesn't seem to overlap exactly with the GTX 2080 vs 1080 graphs. Since this graph doesn't specify which cards are being compared, this could be an apples to oranges comparison.

This shader performance increase must be from INT32 shading, like Jensen says here: https://youtu.be/CT2o_FpNM4g?t=794
Nice to see some games already benefiting from it.

 

[saeedkunna]

is there any news about sli improvements since they are using NVLink it would be interesting to see how t2 2080 would compare to the 2080ti .

 

[IntelUser2000]

There's a new slide claiming shading performance gains vs Pascal.

Disregard anything other than pure performance. Nothing about Ray Tracing performance, Shading performance, performance efficiency, or anything other than just performance.

Modern day marketing teams are guilty about massaging details in such a way that anything they say might as well be a straight up lie. Remember the crap about Polaris being 2.5x better in perf/watt? Or Vega's new features? Yea, ok it was AMD's marketing team, but nearly every marketing team has the same problem.

 

[PeterScott]

Disregard anything other than pure performance. Nothing about Ray Tracing performance, Shading performance, performance efficiency, or anything other than just performance.

Modern day marketing teams are guilty about massaging details in such a way that anything they say might as well be a straight up lie. Remember the crap about Polaris being 2.5x better in perf/watt? Or Vega's new features? Yea, ok it was AMD's marketing team, but nearly every marketing team has the same problem.

Agreed. Everyone should just keep an open mind until third party reviews come in, that is the only real test. Everyone is either quick to praise or condemn (mostly condemn).

 

[IntelUser2000]

Agreed. Everyone should just keep an open mind until third party reviews come in, that is the only real test. Everyone is either quick to praise or condemn (mostly condemn).

I'm on the pessimistic side.

This(https://i.imgur.com/SWMjHOK.jpg) is showing the 2080 is barely better than the 1080 Ti. In Canada, it costs $100-200 more to get the 2080 versus the 1080 Ti.

10 games.

-2 games slower than 1080 Ti
-3 games with large gains(15-30%)
-1 game near 10% faster
-4 games within 5%(up or down)

Average 5.4%. 5% gain in performance for 10-15% increase in prices.

 

[sze5003]

I guess for folks that already have a 1080ti and a 4k monitor don't really need to upgrade. If they are happy with what they are getting in the games they play.

There are a few games coming which I would be interested in playing maxed out namely Hitman 2 since I'm almost done the last one, shadow of tomb raider, resident evil 2 remake, and possibly metro Exodus although I don't remember if I finished the last metro game. I know I have one or two of them in my steam library.

For me the only reason I would have to upgrade is going from 1440p to 4k probably same size monitor. I don't even think it may be worth it as it would be 60hz vs 144hz that I have now because I think the 4k monitors that go to 100hz are way more expensive than I'd want to pay knowing my sporadic gaming habits. Not to mention I hate shopping for components like monitors, such a hassle. Took me forever to decide on the current gsync screen I have now.

And for me RTX is a nice to have not something that's a must have since we won't really get beneficial outcomes from it other then better looking environments but at the cost of a lot of performance and dropping resolution.

I agree that they should just stop with the slow leaks of tiny amounts of info, it doesn't really add anything. It's not like you can go order one right away anyway. But 4k 60fps does sound nice assuming they mean you can run highest settings.

 

[Konan]

I'm on the pessimistic side.

This(https://i.imgur.com/SWMjHOK.jpg) is showing the 2080 is barely better than the 1080 Ti. In Canada, it costs $100-200 more to get the 2080 versus the 1080 Ti.

10 games.

-2 games slower than 1080 Ti
-3 games with large gains(15-30%)
-1 game near 10% faster
-4 games within 5%(up or down)

Average 5.4%. 5% gain in performance for 10-15% increase in prices.

Yes but what resolution are all those FPS scores from on each. Pretty useless to go by right now imo.

 

[IntelUser2000]

Yes but what resolution are all those FPS scores from on each. Pretty useless to go by right now imo.

That's the same number from post #822, where it says 4K HDR.

I'd figure the 50% increase from 1080 to 2080 is right. But 1080 has that 10% penalty with HDR, and if they fixed that with Turing, then in SDR mode the difference might come down to 35%.

35% increase is in line with shader and memory bandwidth increase.

From specs,

2080 Ti: 25-30% faster than 1080 Ti
2080: 25-30% faster than 2080
2070: 35-40% faster than 1070

 

[piesquared]

I'm on the pessimistic side.

This(https://i.imgur.com/SWMjHOK.jpg) is showing the 2080 is barely better than the 1080 Ti. In Canada, it costs $100-200 more to get the 2080 versus the 1080 Ti.

10 games.

-2 games slower than 1080 Ti
-3 games with large gains(15-30%)
-1 game near 10% faster
-4 games within 5%(up or down)

Average 5.4%. 5% gain in performance for 10-15% increase in prices.

Additionally, these are nvidia cherrypicked games with cherrypicked settings. These nvidia marketing slides are nothing but hype train fuel, third party reviews that test additional games outside of nvidias review guidelines will be what shows the real results. Although I expect there will be very few that stray from the guidelines set by nvidia, due to potential repercussions similar to those in the new draconian NDA conditions. I expect HardOCP will be one of the few that show the true performance in a wide variety of games, even if they aren't nvidia sanctioned that show performance when hand tuning isn't or won't be done.

 

[HurleyBird]

This shader performance increase must be from INT32 shading, like Jensen says here: https://youtu.be/CT2o_FpNM4g?t=794
Nice to see some games already benefiting from it.

My assumption is that, at least in current titles, usage of INT32 shading is accomplished via drivers. How much of that is generally applicable vs. explicit to each individual title? Regardless, I can see Turing gaining a performance advantage over other architectures as time progresses, similar to what happened with original GCN.

 

[beginner99]

So just like Vega launch, I will be taking screenshots and showing them to the know it all crowd who just like to baselessly crap on any Turing news for no good reason except AdoredTV and AMD friendlies said so.

And? Given the price hikes anything below 50% increase to their predecessor over multiple game with traditional settings (no hdr or other cheating) is the absolute minimum to even consider these cards. To actually for once get significantly improved performance/$ for a new release from NV they would need to be close to 100% faster. It's clear this won't happen.

 

[Muhammed]

But 1080 has that 10% penalty with HDR, and if they fixed that with Turing,

PUBG, ARK, Tomb Raider, JX3, Wolfenstien 2 and Infiltrator Demo don't have HDR, yet they still show the same 40 to 50% uplift.

This needless crapping on the numbers provided by NVIDIA is pathetic, NV has no reason to lie about them so close to reviews, in fact they never lied about fps uplift with Maxwell or Pascal before.

And after the 50% efficiency increase, the increased bandwidth, and core count, only a fool would claim 10 or 15% increase in performance, even a TitanXP is able to provide that over a 1080Ti.

 

[tamz_msc]

That ~40% better performance in favor of the RTX 2080 against the GTX 1080 in NVIDIA's graphs can be easily explained by a 10% increase due to higher clocks(if it is a FE be FE comparison and if the new FE design is really that much better), increased CUDA core count and higher memory bandwidth.

 

[Muhammed]

To actually for once get significantly improved performance/$ for a new release from NV they would need to be close to 100% faster. It's clear this won't happen.

The price is bad no doubt about that, but the node is new, the die is bigger and GDDR6 is pricier, AND it comes with special eye candy features NOT AVAILABLE ANYWHERE ELSE, so get over it. If you don't like it don't buy it. But to claim performance will be bad after all the data shared is nothing but clear and blatant ignorance.

 

[beginner99]

The price is bad no doubt about that, but the node is new, the die is bigger and GDDR6 is pricier, AND it comes with special eye candy features NOT AVAILABLE ANYWHERE ELSE, so get over it. If you don't like it don't buy it. But to claim performance will be bad after all the data shared is nothing buy clear and blatant ignorance.

Dude, it's about performance/$ and the fact that progress in the whole IT 7 computing sector happened due to getting more performance/$. In the GPU industry this simply isn't true any more. We simply get same performance/$ hence no real advancement because people can't just spend more and more.
This is even worse than the last decade with Intel milking quad-cores. At least Intel kept prices at affordable levels even though performance/$ increase was also lacking.

 

[Muhammed]

Additionally, these are nvidia cherrypicked games with cherrypicked settings.

That's crap, the games cover a wide range of APIs, DX11, DX12, Vulcan, old and new, and both AMD and NVIDIA sponsored titles. If they are cherrypicking they will be doing DX12 and Vulkan only games just like AMD. Or pick only their sponsored titles But guess what? they didn't. This is one of the best wide coverage of games I have seen in a while.

lthough I expect there will be very few that stray from the guidelines set by nvidia, due to potential repercussions similar to those in the new draconian NDA conditions.

LOL so Turing is doomed to be bad even after independent reviews? This is not logic anymore folks, this is just pathertic.