NVIDIA GeForce RTX 2080 Ti Super

[IdiotInCharge]

You probably aren't wrong, but when we're talking about enthusiast class products I'm not really sure things like "efficiency" or "power requirements" really matter all that much. At that level the only thing that really matters is raw performance (and maybe price). If Big Navi if performance competitive with enthusiast class cards then it being less efficient isn't going to matter.

I'd say that the degree varies. Big TDPs aren't a huge issue until they come with big noise levels, or when the GPU is extremely difficult to cool and has a hard time hitting performance numbers due to heat. Or power draw issues as we've seen from a number of AMD GPUs (and the 2080 Ti) in recent memory, where power draw overstresses aging or substandard PSUs.

And the bigger deal is that it's in comparison with what we already have. My overall point is that AMD's current GPU is both slower and less efficient than a competing part that has RT circuitry, which Navi lacks. Just scaling up the same architecture to Big Navi would infer the same relationship and then AMD is adding RT.

I have no doubt that AMD can build a competitively performing part; my doubt lies in whether said part will actually be competitive overall, or if they'll have to bargain-price it out of the gate to get any market traction outside of those that buy AMD because it's AMD or it isn't Intel / Nvidia.

 

[DF-1]

It's too late in the cycle. 7nm or nothing.

 

[Ricky T]

It's too late in the cycle. 7nm or nothing.

Yeah anybody that's waited this long is probably willing to wait longer and just skip this generation entirely. The performance jump over the previous gen and the insane prices just made a lot of 1080ti owners sit tight and I don't think this will change that situation much.

 

[Shoganai]

Yeah anybody that's waited this long is probably willing to wait longer and just skip this generation entirely. The performance jump over the previous gen and the insane prices just made a lot of 1080ti owners sit tight and I don't think this will change that situation much.

I honestly have no idea who they’re targeting with this nonsense.

 

[Snowdog]

Someone posted that in the 3xxx series thread. A 2080ti super is a very boring card... maybe 6% faster than a 2080ti.

If that happens it makes you wonder if nVidia misstepped with Ampere or if it’s an aggressive design.

Personally I’d be happy with RT add in cards. My 2080ti has more than enough rasterized performance.

Certainly not exciting like a new chip, that's blatantly obvious. Especially not exciting if you already have a 2080 Ti.

But what the source of this 6%?? A little boost in clock speed and boost memory to 15.5 GHz like the 2080 Super, should be able net 10%. A nice improvement in bang/buck if the maintain or lower price.

Though this rumor doesn't seem any more credible than it was last time.

Wake me when Videocardz confirms something.

 

[Ricky T]

Certainly not exciting like a new chip, that's blatantly obvious. Especially not exciting if you already have a 2080 Ti.

But what the source of this 6%?? A little boost in clock speed and boost memory to 15.5 GHz like the 2080 Super, should be able net 10%. A nice improvement in bang/buck if the maintain or lower price.

Though this rumor doesn't seem any more credible than it was last time.

Wake me when Videocardz confirms something.

Some of you are being unrealistically optimistic as around 6% is actually about what to expect. Here you can see the Titan RTX is about 5 to 7% faster than the 2080 TI and these are both Founders Editions so apples to apples for coolers.

 

[Nightfire]

Some of you are being unrealistically optimistic as around 6% is actually about what to expect. Here you can see the Titan RTX is about 5 to 7% faster than the 2080 TI and these are both Founders Editions so apples to apples for coolers.

Well, bandwidth is only 9% faster on the Titan, so having only a 6% boost in performance doesn't seem that far off. Witcher 3 saw nearly a 9% boost with that bandwidth so the potential is there. The expectation was to have up to 744 MB/s Bandwidth (20% higher than the RTX 2080ti) versus 616 on the 2080ti and 672 on the Titan RTX.

 

[5150Joker]

Normally I'd agree with that sentiment, except I've experience every AMD GPU release, and well, having little Navi be less efficient without RT really doesn't bode well for Big Navi.

Thing is who cares about efficiency? As long as the performance is there at the right price, that's all that matters. Little Navi put nVidia in enough panic to get it to start releasing all these super cards. Plus it seems AMD is finally gaining marketshare now too. If they put out big Navi by H1 2020 and it gives 2080 Ti levels of performance for $700 but eats 50W more power, who would honestly give a fuck? I sure wouldn't and unless Ampere was set to release around the same time frame, it would be my next gpu.

Here's my bet on big Navi:

Early 2020 release
10-15% faster than 2080 Ti on early drivers
Will eat more power than 2080 Ti
Better DXR performance than nVidia
$700-$750

If the above comes true and 3080 Ti is late, AMD will have me as a customer after more than a decade.

 

[IdiotInCharge]

Thing is who cares about efficiency? As long as the performance is there at the right price, that's all that matters.

It depends.

Little Navi put nVidia in enough panic to get it to start releasing all these super cards.

...not really. Nvidia had those readied in order to lower prices without directly lowering prices on existing retail SKUs. Releasing them before Navi was just good business.

Plus it seems AMD is finally gaining marketshare now too.

In GPUs? Outside of mining warehouses? More like losing marketshare given that they've ignored both the high-end and ray-tracing in products released in the last few years.

If they put out big Navi by H1 2020 and it gives 2080 Ti levels of performance for $700

Notice AMD charging Intel prices for CPUs? You think they're going to price Big Navi that low if they actually have the performance of a 2080Ti?

but eats 50W more power, who would honestly give a fuck? I sure wouldn't and unless Ampere was set to release around the same time frame, it would be my next gpu.

Here's your challenge: if it's just 50w, and the coolers don't suck (or blow!), whatever, at least outside of SFFs and other TDP-limited scenarios (and mobile). However, we're talking about a 2080 Ti 'super' in this thread; if AMD releases Big Navi with the same tech as seen in Navi, and Nvidia releases Ampere on 7nm, either a) Nvidia is going to blow the performance of Big Navi out of the water or b) Big Navi is going to eat over 100w more for similar performance, as Navi at 7nm is already behind Turing at 12nm without ray tracing.

Here's my bet on big Navi:

Early 2020 release maybe
10-15% faster than 2080 Ti on early drivers likely slower
Will eat more power than 2080 Ti without a doubt- hopefully not too much more
Better DXR performance than nVidia highly unlikely -- like Zen, it's going to take developers years to get a handle on Navi RT
$700-$750 if I'm wrong about any of the above, AMDs shareholders would hang the board for pricing so low

My responses in yellow

If the above comes true and 3080 Ti is late

Not only is the above not likely to come true, the 3080 Ti is quite unlikely to be late, just based on how often Nvidia gets stuff out on time and AMD... doesn't.

AMD will have me as a customer after more than a decade.

Pretty sure they've already got you regardless of what they release

 

[5150Joker]

It depends.



...not really. Nvidia had those readied in order to lower prices without directly lowering prices on existing retail SKUs. Releasing them before Navi was just good business.

1660 Super just came out so I wouldn't exactly say NVIDIA had them readied. They may have had a back up plan in case Navi surprised them and it seems it did.

In GPUs? Outside of mining warehouses? More like losing marketshare given that they've ignored both the high-end and ray-tracing in products released in the last few years.

From: https://articles2.marketrealist.com/2019/08/amd-gains-significant-gpu-market-share-from-nvidia/

Looking at the past two instances, we can conclude that AMD’s jump of 940 basis points in its market share is probably due to Navi GPUs. The company shipped its 7nm (nanometer) Navi-based RX 5000 series GPUs ahead of the July 7 launch.

As this is AMD’s next-generation GPU, it might have shipped a larger quantity to meet the initial demand for new GPUs. Another notable point is that AMD succeeded in gaining more than 30% discrete GPU market after excluding crypto-related demand.​

Notice AMD charging Intel prices for CPUs? You think they're going to price Big Navi that low if they actually have the performance of a 2080Ti?

Notice little Navi outperforming their NVIDIA counterparts for less money?

Here's your challenge: if it's just 50w, and the coolers don't suck (or blow!), whatever, at least outside of SFFs and other TDP-limited scenarios (and mobile). However, we're talking about a 2080 Ti 'super' in this thread; if AMD releases Big Navi with the same tech as seen in Navi, and Nvidia releases Ampere on 7nm, either a) Nvidia is going to blow the performance of Big Navi out of the water or b) Big Navi is going to eat over 100w more for similar performance, as Navi at 7nm is already behind Turing at 12nm without ray tracing.

It's a foregone conclusion big Navi won't be just little Navi on 7nm but an updated architecture with RT included with optimizations I'm sure. I think AMD calls it RDNA 2:

Not only is the above not likely to come true, the 3080 Ti is quite unlikely to be late, just based on how often Nvidia gets stuff out on time and AMD... doesn't.

Pretty sure they've already got you regardless of what they release

Haha you think I'm suddenly an AMD fanboy because I think big navi will be a good GPU? Ampere is and always was my first choice but if Big Navi comes out first out of the gate swinging and Ampere is 6+ months out, then yeah, AMD has me. Also yes, NVIDIAs high end prices (and continued uptick each generation) do disgust me so there's that working in AMDs favor. Also I was never a Turing fan, you can check my post history about why I think so--it was basically NVIDIA shitting on its gaming customers by forcing overpriced workstation GPU tech on us to subsidize its data center business.

 

[Legendary Gamer]

Only if they reduce the price as well. The article claims it is targeted at people who still have 1080Tis and don't feel the need to upgrade. That would be me, the 2080Ti was too much money for the improvement offered so I kept the 1080Ti. If the new super card drops down to $800, or maybe $900 then ya I'll probably pick one up. I would have gotten at 2080Ti at $800. However if it stays over $1k? No thanks, I'll keep what I have. New high end GPUs are a fun luxury I like to waste money on, but there are limits to how much I'll waste

You're dreaming. This is Ngreedia we are talking about. They still have 2080Ti's in the channel that are just under a grand to over one... They have no competition on the high end, so they won't release a 2080 Super killer at the same price point.

These cards will be 6-10% faster, will be edging in on a Titan but still gimped enough to warrant the Titan's existence and cost, likely, 1200+ bucks

 

[IdiotInCharge]

1660 Super just came out so I wouldn't exactly say NVIDIA had them readied.

These things take close to a year to ready for release. Nvidia had them planned before Navi was taped out (but was also planned).

Notice little Navi outperforming their NVIDIA counterparts for less money?

Not really?

It's a forgone conclusion big Navi won't be just little Navi on 7nm but an updated architecture with RT included with optimizations I'm sure. I think AMD calls it RDNA 2:

See: every other 'updated' AMD architecture, and now add new, unproven features.

Haha you think I'm suddenly an AMD fanboy because I think big navi will be a good GPU?

Nope, it's the outlandish statements that are completely separate from history, including every time AMD is preparing a new GPU that's expected by their trumpeters to 'own' the competition. Hint: they don't.

Ampere is and always was my first choice but if Big Navi comes out first out of the gate swinging and Ampere is 6+ months out, then yeah, AMD has me.

AMD will come out swinging- we just don't know which weight class they'll be swinging in

 

[IdiotInCharge]

This is Ngreedia we are talking about.

*Corporate America

Wait for AMD to start charging Nvidia prices the very moment they believe they have a slightly faster product.

 

[5150Joker]

These things take close to a year to ready for release. Nvidia had them planned before Navi was taped out (but was also planned).

Where's the proof they take a year to prepare an SKU of an already finished architecture that's in production?

Not really?

Not a good precedent when you got the latest and greatest games showing otherwise:

You've got a 5700 XT basically matching a 2080 at $200 less cost in 2019's hottest release title for the PC. It also beats it's nearest competitor (2070 Super) in this game and still costs nearly $100 less.

Nope, it's the outlandish statements that are completely separate from history, including every time AMD is preparing a new GPU that's expected by their trumpeters to 'own' the competition. Hint: they don't.

You really think it's outlandish to believe that AMD will have Big Navi out on time w/better RT performance than NVIDIA? It's not like NVIDIAs RT performance is exactly a high benchmark to beat. As for release time table, since their console contracts dictate timely schedules, it is why I have some hope Big Navi will show up sooner than later (early 2020).

 

[Legendary Gamer]

*Corporate America

Wait for AMD to start charging Nvidia prices the very moment they believe they have a slightly faster product.

I was too hasty on the naming convention, you are absolutely correct. I was thinking the same thing and had even mentioned in another thread that should AMD begin leading in all segments, expect price hikes to top tier prices (AMD's price hikes have already begun). A market leader with no competition is never a good thing, for us consumers. For the market leader, well, it's where they all want to be. Those that are, fight to stay there. Those that just lost their crown (like Intel) will spend billions to stall the spread of AMD's lead.

 

[5150Joker]

I was too hasty on the naming convention, you are absolutely correct. I was thinking the same thing and had even mentioned in another thread that should AMD begin leading in all segments, expect price hikes to top tier prices. A market leader with no competition is never a good thing, for us consumers. For the market leader, well, it's where they all want to be. Those that are, fight to stay there. Those that just lost their crown (like Intel) will spend billions to stall the spread of AMD's lead.

Yeah anyone that thinks AMD wouldn't charge high end prices if they take the crown are in living in a fantasy world. But conversely, AMD isn't also foolish enough to charge the same prices as Intel/NVIDIA w/out the same level of marketshare + performance. On the CPU front, they still aren't at Intel's performance across the board (e.g. gaming) but I'd say around parity overall but they still charge less than Intel does. As for GPUs, they know they have a lot of work to do and their image with Joe Public isn't near what NVIDIA has established so they will continue to provide better price/performance IMO.

 

[IdiotInCharge]

Where's the proof they take a year to prepare an SKU of an already finished architecture that's in production?

How do you know that it's 'already finished'? The 1660 is a different part.

Not a good precedent when you got the latest and greatest games showing otherwise:

You say 'games', and then you quote one game that was just released. These are different things.

You really think it's outlandish to believe that AMD will have Big Navi out on time w/better RT performance than NVIDIA?

The hardware itself? RT hardware is probably one of the simplest things to design, and AMD are fantastic designers, so no, I don't think the hardware will be slower -- but in AMDs case, that's almost never the problem. Actually putting that hardware to use with software is what trips them up, and that's usually tied not to individual core types in AMDs GPUs being slower, but the balance, communication, and drivers being less capable as a whole.

It's not like NVIDIAs RT performance is exactly a high benchmark to beat.

...it's the only benchmark.

As for release time table, since their console contracts dictate timely schedules, it is why I have some hope Big Navi will show up sooner than later (early 2020).

AMD could have released 'big' gaming GPUs for years, yet haven't done so since they bought ATI. They've instead made big compute GPUs and smaller gaming GPUs, with their big compute GPUs being even less optimized for gaming than Nvidia's. They usually lose out to Nvidia's second or third tier parts in gaming as a result, all while sucking down noticeably more juice, and producing more heat and more noise with their regularly poorly designed coolers.

 

[IdiotInCharge]

Those that just lost their crown (like Intel) will spend billions to stall the spread of AMD's lead.

Advertising, providing discounts, and tuning the hardware that they can produce remind me much of what AMD did... after they released Bulldozer

 

[5150Joker]

How do you know that it's 'already finished'? The 1660 is a different part.



You say 'games', and then you quote one game that was just released. These are different things.

Yes I did, I should have corrected that but still, look at the another hot release for this year that's also DX 12 and you see AMD's 5700 XT matching it's more expensive NVIDIA counterparts just like I said:

99.3 fps vs 101.5 fps (basically indistiguishable) for ~$100 less. The more titles we start seeing with DX 12 as the main API in 2020, I think we'll see the performance gap between NVIDIA and AMD shrink.

The hardware itself? RT hardware is probably one of the simplest things to design, and AMD are fantastic designers, so no, I don't think the hardware will be slower -- but in AMDs case, that's almost never the problem. Actually putting that hardware to use with software is what trips them up, and that's usually tied not to individual core types in AMDs GPUs being slower, but the balance, communication, and drivers being less capable as a whole.

Have we seen any real issues with Navi and drivers? I haven't been aware of any that stand out.

AMD could have released 'big' gaming GPUs for years, yet haven't done so since they bought ATI. They've instead made big compute GPUs and smaller gaming GPUs, with their big compute GPUs being even less optimized for gaming than Nvidia's. They usually lose out to Nvidia's second or third tier parts in gaming as a result, all while sucking down noticeably more juice, and producing more heat and more noise with their regularly poorly designed coolers.

They could have matched Intel for years but didn't either but sometimes corporate directives change.

 

[Legendary Gamer]

Advertising, providing discounts, and tuning the hardware that they can produce remind me much of what AMD did... after they released Bulldozer

The real difference is that AMD didn't have billions of dollars in a slush fund to toss at the problem while they developed their next node and architecture. I really expected AMD to die when Ryzen was introduced, I figured they were pretty much going to be relegated to semi custom business and a low end player like Cyrix/VIA was (before they essentially disappeared).

I am not really knocking what Intel is doing, so much as I am presenting the fact that they will do anything and everything they can to slow Ryzen acceptance and buy time for their next gen stuff. It makes perfect sense.
As for tuning their 14nm+++++++++++++++++++++++++++++++++++++++++++ architecture, it's no longer impressive that they are showing they can squeeze a couple hundred Mhz past 5Ghz out of their products. Not when Kyle is already doing it on a processor 3 revisions behind what they are touting as the next big thing!

 

[Legendary Gamer]

Essentially how I feel about Intel's "tuning" what they have:

and:

 

[Riccochet]

Some of these charts of so far off reality I'm not sure WTF they were using to get the results they did.

A 2080 and 1080Ti trade blows on raster performance. Those charts put the 1080Ti below a Radeon 7 and 5700XT. That's laughable. Especially the COD:MW one. I get 110+ FPS at 1440P Ultra all day. While streaming, at that.

 

[erek]

Essentially how I feel about Intel's "tuning" what they have:

View attachment 198483

and:

View attachment 198484

i'm still waiting for the 9900KFC to be honest

 

[mls1995]

Have we seen any real issues with Navi and drivers? I haven't been aware of any that stand out.

Ever look at the Newegg reviews for 5700 xt models? I did because I was considering one and there were a LOT of driver complaints. I’m honestly afraid to buy a 5700 xt.

 

[Ricky T]

Some of these charts of so far off reality I'm not sure WTF they were using to get the results they did.

A 2080 and 1080Ti trade blows on raster performance. Those charts put the 1080Ti below a Radeon 7 and 5700XT. That's laughable. Especially the COD:MW one. I get 110+ FPS at 1440P Ultra all day. While streaming, at that.

I don't recall the 2080 and 1080ti trading blows as I thought the 2080 always had at least a small advantage even at launch. Over the year and in many newer games, 2080 is starting to put a bigger gap over the 1080ti.

 

[OutOfPhase]

AMD always makes cards which have brutal amounts of raw power. The problem, to use a bad car analogy, is getting it to the rear wheels.

In titles with top shelf developers basically bypassing driver code (DX12/Vulkan) they get incredible results. The question is whether or not that's expected to be the norm, or if those are the exception. I expect AAA devs to have huge teams doing this work well, but we see that writ-large this is not universal. Many companies are still targeting DX11 because low level APIs are a giant pain which require extensive per-card tuning. You can leave it to driver teams, but unfortunately AMD does not do terribly well on that front.

NV is really good at getting their power to the wheels via driver support and developer relations in general. In optimal cases, that brute power AMD has is truly cool. They just need to find ways of getting that power to Development_Firm_004. Asking them to do huge amounts of card tuning via low level drivers simply won't happen. That requires AAA budgets they don't have.

 

[Riccochet]

I don't recall the 2080 and 1080ti trading blows as I thought the 2080 always had at least a small advantage even at launch. Over the year and in many newer games, 2080 is starting to put a bigger gap over the 1080ti.

Barely an advantage. 1-3%. When the 2080 Super dropped even they called it another 1080Ti.

 

[IdiotInCharge]

Have we seen any real issues with Navi and drivers? I haven't been aware of any that stand out.

Well, the main issue is that they're currently lacking ray tracing, and that we don't have games running on AMD GPUs with ray tracing enabled. The only exception to AMDs drivers not being a shitshow has been Vulkan... so I'll upgrade them to a 50/50 there

They could have matched Intel for years but didn't either but sometimes corporate directives change.

They really couldn't have. They committed to Bulldozer when they were truly giving Intel hell and their next architecture was a decade behind.

The real difference is that AMD didn't have billions of dollars in a slush fund to toss at the problem while they developed their next node and architecture.

The biggest problem is that AMD committed to the mistake in the first place, right after Intel finished cleaning up from making the exact same mistake. That's on AMD. If they'd had a competitive product, they'd have continued to give Intel hell and would have made massive inroads into what is now the biggest 'desktop' market: laptops. A space where their products simply don't compete to this day, unfortunately.

I am not really knocking what Intel is doing, so much as I am presenting the fact that they will do anything and everything they can to slow Ryzen acceptance and buy time for their next gen stuff. It makes perfect sense.

That's called business.

 

[Ricky T]

Barely an advantage. 1-3%. When the 2080 Super dropped even they called it another 1080Ti.

I'm looking at the original techpowerup 2080 review right now and it's 10%. And the 2080 super in some newer games is 30 to 40% faster than the 1080ti.

 

[Krenum]

Some of you are being unrealistically optimistic as around 6% is actually about what to expect. Here you can see the Titan RTX is about 5 to 7% faster than the 2080 TI and these are both Founders Editions so apples to apples for coolers.

Twice the money for 2-4fps lol what a jip.

 

[Dayaks]

I don't recall the 2080 and 1080ti trading blows as I thought the 2080 always had at least a small advantage even at launch. Over the year and in many newer games, 2080 is starting to put a bigger gap over the 1080ti.

People may remember that because a lot of them liked to compare heavily OC’d 1080tis to the 2080 FE at the time. It is definitely in the side grade territory regardless. I need 20-30% faster to even notice a difference.

 

[Riccochet]

I'm looking at the original techpowerup 2080 review right now and it's 10%. And the 2080 super in some newer games is 30 to 40% faster than the 1080ti.

 

[Ricky T]

And that doesn't change anything about what I said. You can go to techpowerup yourself and look at exactly what I'm talking about. Heck in Control the 2080 Super is 40% faster than the 1080 ti at 1440p. Essentially the games that are the most demanding are the games where the 2080 Super is actually looking like a decent upgrade. Pretty much all the games where there's not a huge difference, the 1080ti is already at a high frame rate.

https://www.techpowerup.com/review/control-benchmark-test-performance-nvidia-rtx/5.html

 

[honegod]

It is probably silly but driver code for hardware seems to me to be a good use for that newfangled machine learning, a few score Epycs working the problem should be able to sort out all the difficulties in short order.

 

[MangoSeed]

My responses in yellow

My eyes!!

You must be using dark mode. Yellow text is not the business on a white background.

 

[IdiotInCharge]

My eyes!!

You must be using dark mode. Yellow text is not the business on a white background.

What savage would use anything but dark mode?

 

[Snowdog]

I didn't even know there was a "light" mode. Just tried it: Yuk.

Not against white web pages, but this one looks much better dark.

 

[Tsumi]

I like using most websites in dark mode. Much easier on the eyes, especially in the dark.

 

[MangoSeed]

What savage would use anything but dark mode?

Yeah dark mode is far superior in general but I prefer black text on a white background on mobile devices.

 

[PeaKr]

Dark mode performs 10% better than light mode using the latest drivers.