AMD Navi 23 ‘NVIDIA Killer’ GPU Rumored to Support Hardware Ray Tracing, Coming Next Year

[DDH]

It does have a 4K 60hz mode but as tested by digital foundry its dynamic and drops down to 1800p

what is really interesting is that 1080p is way less than half of 1800p, the series X should be able to get 120fps without having to use 1080p if it scaled perfectly . We could have found RDNA2's Achilles hill - it seems to have the same texture througput issues at lower resolutions that Ampere has

Did you even watch the video? They didn't have a hdmi2.1 cable so had to choose a lower resolution. Talk about a **** channel, can't even buy a HDMI 2.1 capable cable to truly review the 4k120 dynamic resolution mode

 

[gpuerrilla]

Lol. Haters gonna hate @DDH

 

[LePhuronn]

https://twitter.com/LisaSu/status/1315648926993854466

It's funny how much people are reading into this tweet. 'as we show off our “Big Navi”' simply refers to a single GPU, or they wouldve said Big Navi graphics cards.

The 2nd part, '@Radeon RX 6000 series !!' could just mean its a part of this series, and they just wanted to link to the Radeon RX page...

At this point, we can safely say there are two designs, the larger variant with 3 fans is big Navi, the smaller one with 2 fans is a less powerful graphics card.

Also, note how confident and up front AMD were with their Zen 3 series announcement this year, there was no need for AMD to be coy with any details.

Um..."show off our 'Big Navi' RX 6000 graphics cards' is a single sentence, that comma is very likely a typo because it makes no grammatical sense. So yes, they will be showing multiple graphics cards at the event. I really don't see how you've convinced yourself there are only 2 cards though. There literally is no evidence to support that, so it's not "clear" in any way.

And there is a big difference between coy and teasing. I really don't understand how people cannot understand this simple concept: October 8th was about Ryzen. RX 6000 was a little teaser to make sure you come back on 28th. Why would AMD reveal anything of substance about their GPUs at the CPU event? If that teaser was literally the very top RX 6000 then it makes 28th event utterly redundant. Why can you not see that?

 

[DDH]

https://twitter.com/LisaSu/status/1315648926993854466

It's funny how much people are reading into this tweet. 'as we show off our “Big Navi”' simply refers to a single GPU, or they wouldve said Big Navi graphics cards.

The 2nd part, '@Radeon RX 6000 series !!' could just mean its a part of this series, and they just wanted to link to the Radeon RX page...

At this point, we can safely say there are two designs, the larger variant with 3 fans is big Navi, the smaller one with 2 fans is a less powerful graphics card.

Also, note how confident and up front AMD were with their Zen 3 series announcement this year, there was no need for AMD to be coy with any details.

Your lack of self awareness is impressive, laughing at people for reading in to a tweet... While also reading in to a tweet

 

[EastCoastHandle]

No I disagree with the characterization that it's "all that is offered" because that's downplaying what is being offered. Yes we agree on the facts because the facts are indisputable, modern games using RTX technology use a hybrid rendering system where most of the scene is rendered using classical rasterization and special ray tracing functions are used for small amounts of the image, normally reflections in puddles or shiny surfaces, and such reflections offer higher quality effects than what you can achieve with rasterization which only does a bad approximation. So that's a benefit to the gamer because we can make these things look better using new techniques.

When it comes to the actual games we play, they don't exist in the context of what ray tracing actually offers when it comes approaching IQ beyond rasterized look. And you are creating a strawman with it's implicity because you cannot argue that it does. Whether, it's your strawman or it's not available in the games we play simply doesn't change the fact that it's not available.

We won't see games that look like that demo be it you want to argue your own point of "all that is offered" or not. It simply won't change what's actual available to games today.

And again your stance on this is somewhat reinforcing what I said about dichotomy.

I don't agree you are applying the term correctly.

When only two choices are presented yet more exist, or a spectrum of possible choices exists between two extremes. False dilemmas are usually characterized by “either this or that” language, but can also be characterized by omissions of choices. Another variety is the false trilemma, which is when three choices are presented when more exist.

I haven't presented you with limited choices. Nor subscribe to your strawman. However, I did bring forth the opinion that, do to the RT Demo's, etc. That ray tracing is very limited to how we see it in the games we play now. I assume you intended to say false dichotomy as I've not heard of just dichotomy in the context you seem to imply. Nor can you accuse someone of being "somewhat" dichotomy. At this point you're simply name calling in an attempt to label.

Technology doesn't just magically appear one day fully formed in all it's glory, completely and totally polished and ready to be used.

That's none of my concern. My position is that even though ray tracing is used in games those games still have a rasterized look that limits how well IQ could be if it was fully RT'd. That hasn't changed.

It's a slow process of refining the tech and rolling it out to users and having them buy into the tech and then iterating on it and making it better over the years until it gets to where we want it. If everyone was like you and dismissed it as a gimmick in the form it is now then it'd never get made because no one would buy into it and thus you wouldn't get that slow incremental advancement that is core to driving all technology growth.

I am fully aware of that. However, this is done at the cost of those who invest in it with nextgen hardware. It's a hard sell, to me, to buy new HW in which an API isn't fully development. So by this statement alone it would certainly cause reservation until I can see there is a level of maturity to it. Which is the gist of my replies on the subject. So apparently we agree.

What you're doing is trying to dismiss this as a gimmick because it's not all 100% ray tracing like the marbles tech demo, therefor why care? And that is what I'm saying is the false dichotomy here, it's not all or nothing, we don't need 100% ray tracing before we see a benefit of it, we can use ray tracing for small parts of the scene and see small fidelity increases.

Again, whether you believe in your own strawman or not what I or someone else disagree with is still a valid opinion to have. However, my opinion is that RT, as demo'd vs what we actually see in games causes a disparity in IQ that cannot be ignored. And because it's secondary to rasterized games does come off as a gimmick. That's the point I'm making. There is no false dichotomy.

And if you want to show what ray tracing actually can do you would need a 100% ray traced environments. Which is why the demos were presented. Unless you contend to imply that one should advertise 100% ray traced demo as a representative to roughly 1/3 of that found in actually games.


IE: RT Demos are showing what rt can really show in IQ.
However, those are just demos and not found in actual games.
Therefore, I am skeptical that RT is beneficial in the game we play now. Even though we have HW for it. At best we only get elements of RT in rasterized games.
Counter Argument: Well, you're just posting a false dichotomy...what did you expect...
Haaa!

 

[g67575]

Your lack of self awareness is impressive, laughing at people for reading in to a tweet... While also reading in to a tweet

I'm just reading the tweet without assuming anything that hasn't been written, in that tweet, or in a previous tweets.

Im sorry you can't accept that sometimes, other people have a different perspective on things

 

[DDH]

I'm just reading the tweet without assuming anything that hasn't been written, in that tweet, or in a previous tweets.

Im sorry you can't accept that sometimes, other people have a different perspective on things

It's funny how much people are reading into this tweet. 'as we show off our “Big Navi”' simply refers to a single GPU, or they wouldve said Big Navi graphics cards.

In your own words you are interpreting what was written, also called reading in to...


But please continue. At least this thread will be entertaining, if not informative, for the next 2 weeks

 

[PrincessFrosty]

When it comes to the actual games we play, they don't exist in the context of what ray tracing actually offers when it comes approaching IQ beyond rasterized look. And you are creating a strawman with it's implicity because you cannot argue that it does. Whether, it's your strawman or it's not available in the games we play simply doesn't change the fact that it's not available.

We won't see games that look like that demo be it you want to argue your own point of "all that is offered" or not. It simply won't change what's actual available to games today.

I'm not really disagreeing with what you're saying here. Ray tracing can be used to varying degrees and right now it's used in small amounts in most of the games it is used in and that will be the case in the upcoming titles, purely because it's too expensive to do in large amounts. You're saying that it's not available in games we play which seems to be a subtle shifting of goalposts here. Who is "we"? Many gamers have been enjoying RT effects in games they play since the RTX 2000 series launch and many are looking forward to RT effects in games soon to be launched.

I don't agree you are applying the term correctly.

I haven't presented you with limited choices. Nor subscribe to your strawman. However, I did bring forth the opinion that, do to the RT Demo's, etc. That ray tracing is very limited to how we see it in the games we play now. I assume you intended to say false dichotomy as I've not heard of just dichotomy in the context you seem to imply. Nor can you accuse someone of being "somewhat" dichotomy. At this point you're simply name calling in an attempt to label.

To be clear when I said false dichotomy I was referring to your presenting 2 choices.
1) Either a game is fully ray traced and it looks better than what you can achieve with rasterization, let's just take the marbles example since you've used that (and I agree it's a good example)
2) Otherwise it's just a gimmick

And I'm simply saying, I disagree. That you can use a bit of ray tracing for certain graphical effects that look better than the rasterized attempts. Once such real world example is screen space reflections that rasterization uses, compared with ray traced reflections. There's known limitations with screen space reflections which rasterized equivalents do not have, and so the rasterized versions look more true to life. Side by side animated comparisons here for example: https://www.nvidia.com/en-gb/geforce/graphics-cards/rtx-2080/

That's none of my concern. My position is that even though ray tracing is used in games those games still have a rasterized look that limits how well IQ could be if it was fully RT'd. That hasn't changed.

This is where things get a bit more muddy. The rendering of those games is a hybrid rendering, most of the scene is rendered with a classical rasterization engine and then certain parts of the scene are overwritten by a raytraced rendering. In this case the rasterization portion is not a limiting factor for the ray traced portion. If you use a ray traced puddle or reflective car and ray tracing is enabled on that piece of geometry then what is reflected in that is ray traced, just as good as quality as it would be if the whole scene was ray traced. Of course the rasterized portions look the same as before, so the image quality of those portions are the same, but the image quality of the RT portions is better.

Obviously if the whole scene were ray traced it would look a lot better but we can't do that because it would grind to a halt with any modern complex 3d game. What I'm saying is that if we can improve just 10% of the screen with ray tracing then that's a winner because we get some improvement. And I'll take some improvement over no improvement.

I am fully aware of that. However, this is done at the cost of those who invest in it with nextgen hardware. It's a hard sell, to me, to buy new HW in which an API isn't fully development. So by this statement alone it would certainly cause reservation until I can see there is a level of maturity to it. Which is the gist of my replies on the subject. So apparently we agree.

Well the APIs are kinda settled, RTX for Nvidia path on this certain is and DirectX have ratified their own DXR variant as well, so the core parts of the technology are done. Originally in 2018 as a windows 10 DirectX update, and then updated in 2020 to v1.1, so the API for this is mature. And so are most of the basic ray tracing functions and effects, the only thing that's not very mature is the hardware. We're limited in what we can use because we don't yet have the speed for it, that's the bottom line. As any technology stack matures the early adopters are the ones who get first access to it, and they have to put up with limited and often janky implementation.

But the point of technology progress is that if you don't start somewhere you do not advance. if we want good ray tracing we first have to do bad ray tracing and then improve it. For example: If we have 800x600 screens and we want 8k screens first we have to advance to 720p, then 1080p, then 1440p, then 4k and then 8k and all the other steps in between. You can't have a 800x600 screen and say well this 1024x768 screen isn't 8k yet, so it's just a gimmick. It's like...sure, it's like only 10% better. But better is better, let's not downplay better.

I'm not saying you should buy into ray tracing right now, it's only on a limited number of games, if you don't want to jump onboard and you're happy with your video card then I'm 100% fine with that, as you have quite rightly pointed out, there's an early adoption penalty for getting in on new tech early, you're basically beta testers. Where my disagreement is that first of all it's a gimmick and that's just not true, ray tracing has been the holy grail of rendering for an insanely long time, anytime we want the best quality possible like in rendered movies or rendered 3d images we use ray tracing, and doing it in real time is nothing short of an insane human feat of engineering. Doing it partially still gives us partial upgrades in quality.

Again, whether you believe in your own strawman or not what you disagree with is still a valid opinion to have. However, my opinion is that RT, as demo'd vs what we actually see in games causes a disparity in IQ that cannot be ignored. And because it's secondary to rasterized games does come off as a gimmick. That's the point I'm making. There is no false dichotomy.

And if you want to show what ray tracing actually can do you would need a 100% ray traced environments. Which is why the demos were presented. Unless you contend to imply that one should advertise 100% ray traced demo as a representative to roughly 1/3 of that found in actually games.

This is tricky because there are many different types of demos:
1) Full scene ray tracing like the marbles demo. And then there's real games that use that kind of full ray tracing like Minecraft and Quake 2.
2) Mixed ray tracing and rasterization such as the Nvidia reflections demo (the star wars looking one). And there also real games that do this such as Battlefield V.

And actually of all the demos so far shown off, the vast majority are mixed. Marbles is the only full scene ray tracing demo I'm personally aware of, and seems to be somewhat of an exception. If you look on the Nvidia demos page all 3 demos for RT are hybrid (Reflections, Atomic Heart and Justice) https://www.nvidia.com/en-gb/geforce/community/demos/

100% of the Nvidia 2000 series launch event was mixed/hybrid: The Sol demo, the infiltrator demo, the shadow of the tomb raider preview, Metro Exodus preview, Assetto Corsa preview, Atomic heart preview and finally the Battlefield V update.

IE: RT Demos are showing what rt can really show in IQ.
However, those are just demos and not found in actual games.
Therefore, I am skeptical that RT is beneficial in the game we play now. Even though we have HW for it. At best we only get elements of RT in rasterized games.
Counter Argument: Well, you're just posting a false dichotomy...what did you expect...
Haaa!

I suspect having now looked again through all the marketing stuff for both the 2000 and 3000 series launch, looked through all the demos, I'd guess that your perception is that most of them were misleading because they're full scene ray traced and we rightfully cannot expect that in real games. And I don't think that's true. The overwhelming examples we have for ray tracing today are mixed/hybrid with rasterization.

This is kinda a long post, I'm trying quite hard not to be antagonistic in my response, if any of this seems offensive or whatever, that's not how I intended it. I'm genuinely trying to work out why our stances are so different from one another on this and I think I've found a few places where we disagree. Anyway it's a good discussion, my labeling of things as false dichotomies is not intended as an insult or personal attack on you, and if it's a straw man it certainly wasn't deliberate.[/quote][/QUOTE]

 

[g67575]

Its interesting how, as we get closer to the rx 6000 series release, there is even more argument /debate, even as more info from AMD becomes available, and we become more informed.

The level of intransigence is definitely higher than I initially expected , although, perhaps not as high as some are used to on these forums.

 

[Poneros]

Did you even watch the video? They didn't have a hdmi2.1 cable so had to choose a lower resolution. Talk about a **** channel, can't even buy a HDMI 2.1 capable cable to truly review the 4k120 dynamic resolution mode

Did you? It was the TV which couldn't do it not because of the cable. All irrelevant anyway, because render resolution was 1440p dynamic, having a TV with 4K 120hz mode wouldn't have changed how it looked. Learn the difference between signal resolution & rendering resolution!

 

[Joxeon]

Even though I won the F5 battle on a 3080 I'm still looking forward to seeing what AMD can bring this time as competition is exactly what we need in the Gpu market.

 

[g67575]

If that teaser was literally the very top RX 6000 then it makes 28th event utterly redundant. Why can you not see that?

Is that such a big problem? Does having separate product launches for CPUs and GPUs somehow make the products any better, or offer and advantage to customers?

Perhaps one of the reasons we have 2 launches, is because they are still making final adjustments to the RX 6000 series (think I read that in an interview somewhere). In my opinion (nothing more), thats why they didn't combine the launches, as they did with the Navi 10 GPUs and most of the Zen 2 lineup.

We also know AMD likes to undercut Nvidia's GPUs, by waiting for NV to release their products first, which they still have the opportunity to do with the RTX 3080.

 

[Hotwired]

Is that such a big problem? Does having separate product launches for CPUs and GPUs somehow make the products any better, or offer and advantage to customers?

The tech reporters had a lot of negative feedback to give last time AMD did a mega reveal of CPU and GPU in one day.

You need to remember that time it was all live so the tech reporters had to go there in person, take a ton of notes and media recordings then immediately do a massive write up on everything at once to get the early clicks.

This time it was all virtual, pre-recorded and as we found out, reporters had access to the presentation before it went live so in theory it should have been vastly easier to handle.

Nevertheless they split it up and if nothing else it keeps the reporters happy to have a good period of time to pre-write up before the presentation and a good period of time to space out their revenue generating videos which are also publicity for AMD.

So that's two parties who benefit for sure. I guess customers gain value from reporters having no excuses of time pressure to rush out a garbage story just to first it. GN had a solid review of the CPU presentation ready to go as soon as the presentation was done.

 

[EastCoastHandle]

To be clear when I said false dichotomy I was referring to your presenting 2 choices.
1) Either a game is fully ray traced and it looks better than what you can achieve with rasterization, let's just take the marbles example since you've used that (and I agree it's a good example)
2) Otherwise it's just a gimmick

This is your strawman and you contradict yourself and refute your strawman when you say:

I'm not saying you should buy into ray tracing right now, it's only on a limited number of games, if you don't want to jump onboard and you're happy with your video card then I'm 100% fine with that, as you have quite rightly pointed out, there's an early adoption penalty for getting in on new tech early, you're basically beta testers.

I suspect having now looked again through all the marketing stuff for both the 2000 and 3000 series launch, looked through all the demos, I'd guess that your perception is that most of them were misleading because they're full scene ray traced and we rightfully cannot expect that in real games.

It's clear you completely understood my view on the subject (2 pages ago, lol) but have intentionally skewed the conversation, tunneling it through your strawman to draw an alt. narrative. Those 2 choices are yours and yours alone. Everything else after that is now moot at best.

And I'm simply saying, I disagree. That you can use a bit of ray tracing for certain graphical effects that look better than the rasterized attempts

And, I will disagree. As those attempts still produce a rasterized image. It's very difficult to disagree when those actual games will never look like the RT demo counterparts when only a few rt elements are used.

This is where things get a bit more muddy. The rendering of those games is a hybrid rendering, most of the scene is rendered with a classical rasterization engine and then certain parts of the scene are overwritten by a raytraced rendering. In this case the rasterization portion is not a limiting factor for the ray traced portion. If you use a ray traced puddle or reflective car and ray tracing is enabled on that piece of geometry then what is reflected in that is ray traced, just as good as quality as it would be if the whole scene was ray traced. Of course the rasterized portions look the same as before, so the image quality of those portions are the same, but the image quality of the RT portions is better.

This is not relevant to the discussion. Again, the final product is more important then the technique used to produce it.

Obviously if the whole scene were ray traced it would look a lot better but we can't do that because it would grind to a halt with any modern complex 3d game. What I'm saying is that if we can improve just 10% of the screen with ray tracing then that's a winner because we get some improvement. And I'll take some improvement over no improvement.

Again, this contradicts your strawman. That's exactly why we will only see demos of what ray tracing can do. This explains why games will remain rasterized with only elements of ray tracing in them. That 10% (as you claim) is the gimmick when you compare it to the actual demos showing what ray tracing can do. You see these contradictions you are making? Oh course...

Well the APIs are kinda settled, RTX for Nvidia path on this certain is and DirectX have ratified their own DXR variant as well, so the core parts of the technology are done. Originally in 2018 as a windows 10 DirectX update, and then updated in 2020 to v1.1, so the API for this is mature. And so are most of the basic ray tracing functions and effects, the only thing that's not very mature is the hardware. We're limited in what we can use because we don't yet have the speed for it, that's the bottom line. As any technology stack matures the early adopters are the ones who get first access to it, and they have to put up with limited and often janky implementation.

Again, not relevant to the conversation. Again, the end results is what stands here. And as it stands there is no revolution to gaming on the PC as it still remains rasterized.

But the point of technology progress is that if you don't start somewhere you do not advance. if we want good ray tracing we first have to do bad ray tracing and then improve it.

Not on my dime. If that's the case simply keep ray tracing in house until it's at a point where you aren't telling customers that it's "starting somewhere" and have reached your destination. Imagine being told that by a car saleman while you are signing your name on the dotted line for your self driving car. "If we want good self driving car we first have to do bad self driving and then improve it."
Haha
Needless to say but no thank you...

I'm not saying you should buy into ray tracing right now, it's only on a limited number of games...

Oh really? The same could be said with the self driving car. The point being is that it's silly to expect someone to agree with you that buying into/investing into only part of product's feature is a feasible idea. Then take the position to be up at arms about why someone is skeptical about it. This is the whole crux of your argument which you've back peddled on.

This is kinda a long post, I'm trying quite hard not to be antagonistic in my response, if any of this seems offensive or whatever, that's not how I intended it..

Your replies to me have been, at best, incensed and vehement because I used the term "gimmick". I think you've let this get to you more then it should. It's apparent that we will only agree to disagree. Although we exchanged view points I've found most of your replies simply trying to change the narrative my post just to fit your own. In the end it will not change anything.

I think it's a gimmick as it limited in use cases which still renders the image in a rasterized way. After 2 pages of this back and forth that won't change. Not because I'm being stubborn but because it's the absolute truth of those games with only elements of rt..

Moving forward we will see more games incorporating RT elements We will see how that fairs with RDNA 2, Consoles and Ampere in the days to come. But here’s another key point: for those that cannot use rt or not pressed with it will simply disable it because that option will be available. Remember rt elements in a pc game so far is just an option not an evolution in gaming on the pc.

What I see the most challenging though is those open world, sandbox games that use RT elements. To me, that's the key of how well those RT games will be optimized vs non sandbox games.

 

[Allnamestaken]

Lmfao I was not expecting a debate when I opened this thread.

 

[Troezar]

Lmfao I was not expecting a debate when I opened this thread.

There's also a couple of dissertations to read too

 

[Rroff]

Obviously if the whole scene were ray traced it would look a lot better but we can't do that because it would grind to a halt with any modern complex 3d game. What I'm saying is that if we can improve just 10% of the screen with ray tracing then that's a winner because we get some improvement. And I'll take some improvement over no improvement.

This is where path tracing comes in - it isn't unduly troubled by polygon count and many other features which are part of scene complexity and can bring the benefits of the fundamental advancements of ray tracing to games with results that are passably close to a proper full screen, full scene complex ray tracer for gaming purposes. It brings the reality much closer much quicker IMO than people think - sure you have to be a bit selective on how many of the sources of light actually do have a reasonable degree of additional light bounces, etc. (ideally you need around 4x the ray budget compared to Quake 2 at 60+ FPS to avoid noticeable artefacts from the limits of the denoiser and the way light is sampled over multiple frames for performance reasons).

People don't realise that the path tracer in Quake 2 doesn't take advantage of the low geometry and other low complexity of the engine as the Minecraft one does, which is often assumed by the doubters, if you scale up the scene complexity by say "64 times" (high resolution meshes, more shader effects, etc.) you only get a ~4% drop in performance. If you increase the complexity of the path tracer itself you will see a bigger performance impact but that is another matter.

A lot of the problem right now is the lack of people with ray tracing hardware - the token effects in many games come with much of the overhead of a fuller ray tracing implementation but have to remain as token effects so they don't affect the ability to simply turn them off for people without the hardware and many of the ways traditional rendering fakes up graphic features conflicts with the wider use of ray tracing and most developers won't go to the time and resources to maintain two branches of the game or having a plainer looking game without ray tracing.

 

[Richdog]

There's also a couple of dissertations to read too

They are definitely not worth reading, based on previous experience ploughing through posts like that. Lots of blah mixed with more blah with a sprinkle of blah added for good measure. Actual content worth reading could fit into below 150 words.

 

[Nablaoperator]

Overall, Big Navi (80 CU) 30-65% slower than RTX 3080 at RT, depending on nvidia's implementation.
Midpoint 53% slower at RT.

k. lets run few pricing scenarios for Big Navi..
Big Navi will not be DLSS/8k capable.
All scenarios in 4k

• Matches RTX 3080 rasterization
  • 30% slower than RTX 3080 in RT: $599
  • 40-50% slower: $549
  • 60%: $499

• 5% faster than RTX3080 in rasterization
  • 30% slower than RTX 3080 in RT: $649
  • 40-50% slower: $599
  • 60%: $549

• 10-15% faster than RTX 3080 in rasterization
  • 30% slower than RTX 3080 in RT: $749-799
  • 40-50% slower: $699
  • 60%: $649

What do you think?

 

[KentMan]

k. lets run few pricing scenarios for Big Navi..

• Matches RTX 3080 rasterization
  • 30% slower than RTX 3080 in RT: $599
  • 40-50% slower: $549
  • 60%: $499

• 5% faster than RTX3080 in rasterization
  • 30% slower than RTX 3080 in RT: $629
  • 40-50% slower: $579
  • 60%: $529

• 10-15% faster than RTX 3080 in rasterization
  • 30% slower than RTX 3080 in RT: $699- 1200 as this is 3090 territory
  • 40-50% slower: $679
  • 60%: $649

What do you think?

Edited in yellow what i think the prices would be in that scenario, arguably DLSS is a bigger deal than Ray Tracing for many users, and if AMD does not have a DLSS solution, they have no right charging Nvidia prices, as people simply will not accept high prices just on Rasterization performance now. To sway people away from Nvidia they either have to beat them on price, performance and features... its not going to change must just having the fastest card unless its hugely faster for a lot less money, its been seen time and time again people simply will not swap to AMD hardware once they are locked into the the Nvidia ecosystem. We may not like it, but Nvidia did not get their marketshare by accident, year on year consistent performance, better features, more refined and finished products most of the time, or simply paying game devs and working with them to implement their blackbox features.

AMD need a Ryzen moment in the GPU landcscape for it to change on any serious scale, they need RDNA2 or whatever comes next to not only be better, but massively cheaper and offer similar features, as if they do not, its too easy for current Nvidia owners locked in to say "no thanks i'll miss xxxx" if you can cover all the bases and come in at an attractive price, people will change, it happened with Ryzen, they know this, we know this.