David Wang From AMD Confirms That There Will Eventually Be an Answer to DirectX Raytracing

[bug]

What development? What possible influence?

RTX is vaporware, and in fact Nvidia's RT is Hybrid - and therefore just another Gamesworks-like effect bolted onto a conventional rasterization system. Most screenshots have extra faked lighting to make the graphical difference noticeable too lol. Furthermore AMD has no way of affecting a damn thing in development unless they get a new line-up out the door that can compete in framerates across the board.

In other words, AMD can't influence anything until they reboot Radeon first, and there is nothing to add input to anyways because RTX is vaporware.

My mistake, I didn't pay attention who I was replying to.

 

[Vayra86]

Well, I did study a bit of computer graphics in college and I know this is a game changer. At the same time, I can understand for the uninformed this may seem like a fad (and the risk of it not catching on in this current incarnation is there). But ray tracing is something that will happen. And I hope it happens as soon as possible.

I do believe that is true. It will happen. But not because Nvidia says so. There is no industry buzz, no support, no movement to speak of. RTX just appeared overnight and suddenly we are in a new realm of ray tracing goodness? That is not how it works. The support from publishers and devs is telling, there isn't any, except those who got paid.

When it comes to these kinds of crystal ball things, I tend to look at the economy of things to get a handle on it. And economically, Turing as an architecture really doesn't seem viable. How are they ever going to scale this up? Through node shrinks? That is going to be short lived. A whoppin' 18-20% of the die is now reserved for RT capability and the performance is going nowhere. Another economical factor is the state of the marketplace, and this very article underlines how sad it is. AMD is not even in a position to make a meaningful entrance with RT enabled GPUs anytime soon. So in terms of market share and adoption we have literally three (!) GPU models capable and even the strongest one has nothing to show for it.

 

[lexluthermiester]

Nevertheless, this executive seems to be that "AMD will definitely respond to DXR"

Saw this coming..

The rest of what he said was a bit disappointing though..

 

[jigar2speed]

"the spread of Ray-Tracing's game will not go unless the GPU will be able to use Ray-Tracing in all ranges from low end to high end"

its upto the GPU manufactures(AMD,Nvidia) that need to develop GPU that can optimise this tech from Low-end to High-End.
who is he telling the point to??...it is they(AMD) that must bring it to all consumers by collaborating with game developers & building hardware.. not wait for ray a light from God to do that..

Fatal fury on my brain... Please stop.

 

[lexluthermiester]

There is no industry buzz, no support, no movement to speak of.

What are you talking about? Talk of RTRT is all over the place. There are over 100 games being made at this point which will have RTRT on release. In august that list was just 20ish. It's a brand new technology. The uptake is always slow. Look at Vulkan and DX12. Talk about slow uptake. DX12 has been out with Win10 for how long? And there's barely a selection of games, 3 years later. Vulkan? Very similar. Most games are still being made with/for older API's in mind. RTRT will be no different. Games will be released as devs learn and demand grows. And that demand is well established with RTX still flying off shelves as fast as they can make them.

 

[FordGT90Concept]

Shadow Warrior 3 confirmed? They got the Wang already

"This is my rifle, this is my Wang! Wait a second."

I love the wording of it. "Eventually"... different spelling for 'not anytime soon'.

It's going to be a good 5+ years before RTRT isn't niche. Probably much longer, especially if Moore's Law continues to be as dead as it appears to be. I mean, Unreal Engine and Unity Engine still don't have proper Vulkan/DirectX 12 support. How long is it going to be before they bother to implement DirectX Raytracing? A decade? More? Not to mention, game development is a constant struggle against low framerates no matter the engine. DXRT increases that problem expontentially.

AMD shouldn't prioritize it. It would be a terrible business decision. They already have Radeon Rays for professional situations which can be pre-rendered.

 

[bug]

I do believe that is true. It will happen. But not because Nvidia says so. There is no industry buzz, no support, no movement to speak of. RTX just appeared overnight and suddenly we are in a new realm of ray tracing goodness? That is not how it works. The support from publishers and devs is telling, there isn't any, except those who got paid.

When it comes to these kinds of crystal ball things, I tend to look at the economy of things to get a handle on it. And economically, Turing as an architecture really doesn't seem viable. How are they ever going to scale this up? Through node shrinks? That is going to be short lived. A whoppin' 18-20% of the die is now reserved for RT capability and the performance is going nowhere. Another economical factor is the state of the marketplace, and this very article underlines how sad it is. AMD is not even in a position to make a meaningful entrance with RT enabled GPUs anytime soon. So in terms of market share and adoption we have literally three (!) GPU models capable and even the strongest one has nothing to show for it.

Yes, I also see Turing as a technology demonstrator (and am quite puzzled it wasn't released as a Quadro-only part). Those prices in consumer space are something I hope never to see again.
What I wouldn't want to see is another TruForm that took 8 years to catch on as Tessellation. And to this day, the hardware still chokes on it.

 

[hat]

Did AMD and nVidia trade places? Usually AMD is the one coming out with all this new tech that nobody uses... until someone eventually uses it and it's not even called what it was originally. Or maybe nVidia just pushed out another turd that will be present in a few things only to be replaced by something else that's much more widely available and works better.

 

[FordGT90Concept]

What is GameWorks? Both have a long history of creating APIs with the express purpose of increasing market share.

 

[Vayra86]

Yes, I also see Turing as a technology demonstrator (and am quite puzzled it wasn't released as a Quadro-only part). Those prices in consumer space are something I hope never to see again.
What I wouldn't want to see is another TruForm that took 8 years to catch on as Tessellation. And to this day, the hardware still chokes on it.

I think RTRT is in a far worse place than Tesselation if you look at the die space it takes in Turing at this moment.

What are you talking about? Talk of RTRT is all over the place. There are over 100 games being made at this point which will have RTRT on release. In august that list was just 20ish. It's a brand new technology. The uptake is always slow. Look at Vulkan and DX12. Talk about slow uptake. DX12 has been out with Win10 for how long? And there's barely a selection of games, 3 years later. Vulkan? Very similar. Most games are still being made with/for older API's in mind. RTRT will be no different. Games will be released as devs learn and demand grows. And that demand is well established with RTX still flying off shelves as fast as they can make them.

'Talk is all over the place'. Where? I'm not seeing it. Some Nvidia press releases and a message about BFV not being ray traced heaven after all, abysmal performance and 'over a 100 games being made'... lol. News flash, devs don't make 100 games for a tiny bit of the marketplace. They add on some afterthought ray tracing to please Nvidia's bag of money. We haven't seen any live manipulation of ray traced content, I've said this before, there is literally NO example to be had of ray tracing making an actual difference for a game's concept. There wasn't one prior to launch, not at launch during the keynote, and there isn't one to be found post launch halfway November.

You've said it yourself. There was more 'talk' about how DX12 and Vulkan were going to revolutionize gaming and look where we are today. Marginal, if even that, performance wins, and not a single game is out that truly leverages the feature levels DX12 has to offer. When Mantle was a thing, we got served with special demos showing what increased draw calls could mean for gaming. Where are the games using that? We have Ashes... and that's it.

The vast majority of games still gets limited by DX11 because the vast majority of the marketplace is still optimized for that API. RTX is in far worse place right now in every possible way. Like bug said, this could well take a decade and if there is one certainty, it is that the current RTX implementation is dead on arrival.

 

[bug]

I think RTRT is in a far worse place than Tesselation if you look at the die space it takes in Turing at this moment.

How much die space did TruForm need?

 

[Vayra86]

How much die space did TruForm need?

https://www.techpowerup.com/gpu-specs/radeon-8500.c93

I couldn't tell you honestly. But let's put them side by side anyway. Put those highlighted bits next to Turing and you can see my point

The problem I think is clear. RTX is directly eating into the remaining headroom and die space we so desperately need to scale performance. TruForm was just bonus as there were tons of options to scale further.

 

[lexluthermiester]

'Talk is all over the place'. Where? I'm not seeing it.

What are we all talking about? And this is happening all over the place.

News flash, devs don't make 100 games for a tiny bit of the marketplace.

Correct. They add RTRT code to their projects for the audience that has RTX to take advantage of. And that "tiny marketplace" is growing by leaps and bounds. That's millions of cards sold so far and millions more for the holiday season. RTX is here to stay and it's not going anywhere.

When Mantle was a thing

Mantle became Vulkan. And there's arguably more support for it than for DX12. But it seems we can agree, adoption takes time. There is one big difference between DX12, Mantle, Vulkan and RTRT; RTRT is a lighting feature set not an API. RTX's version of RTRT can be worked into any existing API, including if someone wanted too, older versions of DirectX such as 11, 10, 9.0c and even 8 if someone got creative. OGL would be a breeze.

RTRT is a platform agnostic feature set..

RTX is in far worse place right now in every possible way.

Not really and for reasons stated above.

Like bug said, this could well take a decade and if there is one certainty, it is that the current RTX implementation is dead on arrival.

Can't agree with that perspective. It's not going to take 10 years. The reason DX12 wasn't jumped to by devs is because a large portion of the Windows gaming market refused to jump on the Win 10 bandwagon. According to Steam a majority still haven't. But RTX/RTRT is, as stated above, platform agnostic. It can be done on any OS with any API.

RTRT is the future of lighting in games. It's not a fad and it's not going away. It's here to stay. Get on board, or join the DoDo bird..

 

[Vayra86]

What are we all talking about? And this is happening all over the place.

Correct. They add RTRT code to their projects for the audience that has RTX to take advantage of. And that "tiny marketplace" is growing by leaps and bounds. That's millions of cards sold so far and millions more for the holiday season. RTX is here to stay and it's not going anywhere.

Mantle became Vulkan. And there's arguably more support for it than for DX12. But it seems we can agree, adoption takes time. There is one big difference between DX12, Mantle, Vulkan and RTRT; RTRT is a lighting feature set not an API. RTX's version of RTRT can be worked into any existing API, including if someone wanted too, older versions of DirectX such as 11, 10, 9.0c and even 8 if someone got creative. OGL would be a breeze.

RTRT is a platform agnostic feature set..

Not really and for reasons stated above.

Can't agree with that perspective. It's not going to take 10 years. The reason DX12 wasn't jumped to by devs is because a large portion of the Windows gaming market refused to jump on the Win 10 bandwagon. According to Steam a majority still haven't. But RTX/RTRT is, as stated above, platform agnostic. It can be done on any OS with any API.

RTRT is the future of lighting in games. It's not a fad and it's not going away. It's here to stay. Get on board, or join the DoDo bird..

I see a trend with you, you always overhype new stuff for yourself and then you're surprised people don't jump on the bandwagon. How's that Atari console coming along... I'm a more realistic type of guy:

RT is the future, VR is the future, low level APIs are the future. Its all the future, and I can't disagree on that. But 'future' is a very broad term. All you need to do is look at history and you can see that these changes are very gradual and some of them are so gradual, you can easily ignore them. I think RT is one of those. We're already looking at 2020 or later at the *very least* to see, *maybe*, the upper half of the GPU product stack as RT capable - with only the top parts actually playable. That is two years essentially lost in terms of adoption rates.

As for those millions of GPUs being sold, where do you get these numbers? Answer: you don't, its a figment of your imagination. The amount of benches ran on RTX GPUs is incredibly low at this point, and its market share is lower than the adoption of 4K monitors right now.

We are now looking at a whoppin'...

Wooops. 9K samples across all RTX enabled cards. Of course, it will rise. But this doesn't look like something that is flying off the shelves, to me. This includes all those pre orders.

Take note also of the ratings: 56 ~ 65. Another showcase of 'hype' to you? Of millions of sales? Of Nvidia that can't make enough RTX to feed this ray-tracing starved world? This echoes the feedback RTX has gathered since it launched. Even in the back of a buyer's head there is that nagging question: "Nvidia, wtf are you doing?"

Some perspective, as well. Take note of market share per GPU tier (of course, high end is more heavily represented here, people are more likely to bench it), but especially take note of the user ratings.

 

[bug]

I see a trend with you, you always overhype new stuff for yourself and then you're surprised people don't jump on the bandwagon.

That's what us tech enthusiasts do: we get enthused about new tech. We like new tech. We like reading about new tech when we can't afford it.
Are we not allowed to?

 

[Vayra86]

That's what us tech enthusiasts do: we get enthused about new tech. We like new tech. We like reading about new tech when we can't afford it.
Are we not allowed to?

Never said anyone's not allowed to do anything, but to then proclaim the next best thing as a universal truth when all the numbers and market sentiment shows something else, just seems to me like having blinders on. It happened with VR. It happened with APIs. It will happen with RTX.

Enthusiasm is abused to sell you overpriced, mostly worthless junk, that is the whole business model of the RTX family. There isn't a single example in GPU history that shows otherwise. If you want to early adopt this crap, be my guest, but wise it surely is not, and in some ways even damaging towards the end goal: nice games. RT introduces expenses that do little to elevate gameplay and money can only be spent once - both for publishers and devs as for consumers.

 

[bug]

Never said anyone's not allowed to do anything, but to then proclaim the next best thing as a universal truth when all the numbers and market sentiment shows something else, just seems to me like having blinders on. It happened with VR. It happened with APIs. It will happen with RTX.

And proclaiming it as the next fail is the right thing to do? Surely you realize how meaningless this discussion is.

Enthusiasm is abused to sell you overpriced, mostly worthless junk, that is the whole business model of the RTX family. There isn't a single example in GPU history that shows otherwise. If you want to early adopt this crap, be my guest, but wise it surely is not, and in some ways even damaging towards the end goal: nice games. RT introduces expenses that do little to elevate gameplay and money can only be spent once - both for publishers and devs as for consumers.

Yes, there's the early adopter tax. I have rarely (if ever) paid it. But faulting those that chose to is just idiotic. After all, if everybody would strictly buy when the price is right, we'd never have new stuff.

 

[jabbadap]

Never said anyone's not allowed to do anything, but to then proclaim the next best thing as a universal truth when all the numbers and market sentiment shows something else, just seems to me like having blinders on. It happened with VR. It happened with APIs. It will happen with RTX.

Enthusiasm is abused to sell you overpriced, mostly worthless junk, that is the whole business model of the RTX family. There isn't a single example in GPU history that shows otherwise. If you want to early adopt this crap, be my guest, but wise it surely is not, and in some ways even damaging towards the end goal: nice games. RT introduces expenses that do little to elevate gameplay and money can only be spent once - both for publishers and devs as for consumers.

Nah RTX will go on with Optix. Real time ray tracing in 3d designing is one of the most wanted aspect in pro world. RT for gaming is just a by-product of that and yeah I agree there's hefty early adopters tax on it. But one die shrink later and prices should normalize.

 

[Vayra86]

And proclaiming it as the next fail is the right thing to do? Surely you realize how meaningless this discussion is.

As meaningless as just about every other discussion here, let's face it... We discuss this to bounce ideas and beliefs off one another, and I try to offer perspective to support those ideas in my posts on the subject. Such is the nature of speculation...

There is some value to be had in figuring out what is going to work well and what is not. What bandwagon is worth jumping on and which one is not. Blindly early adopting everything because some company put a nice sticker on repurposed datacenter technology is... well. You know. And I hate to sound arrogant, but my hit rate is well above 50% when it comes to these predictions. Maybe I should make a video, so people can link it here and it becomes credible...

 

[lexluthermiester]

I see a trend with you, you always overhype new stuff for yourself and then you're surprised people don't jump on the bandwagon.

I'm an optimist. And I'm a much happier person than most.

How's that Atari console coming along...

Doing very well. They raised $3million+ which was 2900% of their goal with a scheduled release of June 2019.

I'm a more realistic type of guy

Can't see that myself. To me at least, you seem on the pessimist side of things. No offense intended.

RT is the future, VR is the future, low level APIs are the future. Its all the future, and I can't disagree on that.

Good. Let's agree to disagree on the rest of the details until more things have happened.

Enthusiasm is abused to sell you overpriced, mostly worthless junk, that is the whole business model of the RTX family.

On this point I will chime in. I own a 2080 and love it! Yes it was pricey, but it performs like a beast compared to my 1080 and I am happy with it. Looking forward to new games that use RTRT and mods for older games. Would so love to see Return To Castle Wolfenstein with RTRT worked into the RealisticHD or Venom mods.

On the point of optimism, I can't wait to see AMD's efforts in the area of RTRT. Hopefully they get the lead out sometime soon.

 

[bug]

As meaningless as just about every other discussion here, let's face it... We discuss this to bounce ideas and beliefs off one another, and I try to offer perspective to support those ideas in my posts on the subject. Such is the nature of speculation...

There is some value to be had in figuring out what is going to work well and what is not. What bandwagon is worth jumping on and which one is not. Blindly early adopting everything because some company put a nice sticker on repurposed datacenter technology is... well. You know. And I hate to sound arrogant, but my hit rate is well above 50% when it comes to these predictions. Maybe I should make a video, so people can link it here and it becomes credible...

As I have said, I'm not big on paying early adopter's tax either. But RTRT is something I'm really excited about. Turing is not going to take the world by storm, but I really believe Nvidia has a plan for it that stretches well beyond Turing.
You want to paint RTRT as a fad, it's your right to do so.
And if we're talking hit rates, I also didn't buy into VR (or 3D before it) and told people here expecting everything to go DX12/Vulkan is like expecting all programmers to give up higher-level languages and start writing everything in C.

 

[lexluthermiester]

Turing is not going to take the world by storm, but I really believe Nvidia has a plan for it that stretches well beyond Turing.

Exactly. NVidia and even AMD are thinking long term. This is both a good thing but also very healthy from an economic point of view.

 

[Vayra86]

Oh my god, I sense consensus between us! This is getting dangerous now, I'm out.

 

[ZoneDymo]

This man gets it.

We heard that even Nvidia's flagship cards buckle hard under their advertised ability of ray tracing (30fps 1080p...).
So that is just going to be nothing more then a silly gimmick (like PhysX) until there is proper support for it.

 

[Vya Domus]

There are over 100 games being made at this point which will have RTRT on release.

No there aren't. Where did you get that from ?