Impact of nVidia Turing RayTracing enhanced GPUs on next-gen consoles *spawn

You keep on using that word. I do not think it means what you think it means.

Once again, no one here is in opposition of Ray Tracing. Stop being paranoid and thinking everyone is against it. What we are here is discussing what is the equation of cost to benefit ratio.

 

Indeed, allow me to rephrase: the opposition of ray tracing in the next console cycle on the grounds of it's practicality and cost.

 

With that said, I wish we could jump forward 1 year and see what RTX^2, or whatever the next Nvidia release is called, brings to the table and see what response, if any, Intel, AMD, and even Apple brings. I'm impatient and it seems like we have a lot of waiting to see how the field is staged. Hopefully the developers will learn a lot and share just as much with everyone during this time and not hide it away behind closed doors.

If next-gen consoles do not have Ray-Tracing technology for 2020 what should Sony and Microsoft do? Should they redesign and delay release until 2021 to include it? Should they move forward with plans for 'standard rasterization' in 2020 and rapid release next-next-gen Ray-Tracing Console in 2023? I don't mean a "mid-gen" console in 2023 like 4Pro or OneX either.

 

From the Eurogamer/DF article: Battlefield 5's RTX ray tracing tested: is this the next level in gaming graphics?

This quote should shock you for several reasons.

• DXR was announced Mar 19, 2018 (DirectX Developer Blog) - but last year puts us in 2017
• By Mar 19, 2018 Nvidia's tools for DXR were then available to the masses for exploration and development

We should be asking how long the DX team has been working on this for. More importantly if DICE would be first to know about this future pipeline before Microsoft AAA studios. I doubt that very much.

• For 343i Halo 5 released Oct 27, 2015 and they are building an entirely new engine from scratch for this next title; they want this next title to deliver for xbox, so expect a tour de force here. My post on evidence of ray tracing in Halo Infinite here
• Turn 10 has no Forza 8 for 2019
• The only expected release anytime soon is Gears 5.
• So it wouldn't surprise me if 2 of 3 of their major AAA studios are prepping to launch with HRT for next gen. Except that UE4 has RT already in the engine and is likely to continue development.

I don't know the cost structure, nor do I know how it will work. I don't know the vendor or the price. But MS has tied this to DX12 and invariably Windows 10. There's no doubt in my mind strategically you want the console to help push DXR thus the rest of the market off Windows 7 and onto 10. Would MS really go out of their way to sabotage their own success with DXR? Did Phil Spencer really do all of this pivoting to save Xbox and start making it successful again only to push another 4K box? That's certainly not going to get more people to switch away from PS5. That's certainly not going to get more people onto Windows 10. Microsoft's greatest strength is technology, so if they aren't deploying new technology they are falling behind their competitors.

Primitive shaders are all things their competitors have access to, but RT acceleration there's no guarantee with AMD as we understand it now. Nintendo could have access to RT as early as their next generation after NSW! That would be a rough day for MS and Sony if they launched next gen without it.

So I respect all the work being done to estimate feasibility, but we need more information before we can start guessing further. We're just too far out. 2 years is a lot of time for things to change. As we know, the difference between PS4 having 4 or 8 GB of memory was less than a year. The difference between Xbox One wasting 30% of its silicon on esram vs having 8 GB GDDR5 is 1 year.

Imagine 2 years time from now, or if you'd like 3+ years for MS AAA studios working with RT technologies. Xbox BC hardware was brought into Xbox One just a year before launch. Xbox has GPU-driven dispatch as well. So the hardware can definitely move in-front of the software, and now we're seeing the software definitely move in-front of the hardware.

I don't think anyone has said anything that is remotely close to the final possibilities except for the price points -- which will lie between 399 and 499.

Lots of time for things to change in favour of DXR. Things certainly aren't going to get worse.

 

Never underestimate AMD

 

"Stupid console gamers". Please don't put words in my mouth.

Also, if it's about console gamers, why would the opinion of PC gamers obessed with performance be relevant? We know console gamers are fine with high end games running at 30fps.

Oh, so it's all about dismissing opinions you don't agree with. Crysis was absolutely beyond what was possible on consoles. And yes, that means performance suffered. THAT's the point. Even Crytek's own attempt at porting the game to them with a revamped engine had noticeable compromises (like the lack of volumetric lighting/shadows).

It is a fact that after Crysis (and the STALKER games), high end PC gaming consisted pretty much of only console ports with better settings, nothing designed with major advances in technology. Sorry if that pisses you off.

Consoles are years away and your jumping to conclusions based a rushed, tacked on implementation on a hardware architecture NOT designed by AMD ? That seems very reasonable.

 

What conclusions? Again, putting words in my mouth (and everyone else’s).

Brit articulated my exact thoughts best:

We are all impatient to see what happens, as no one can come to a “conclusion” yet.

 

We need info about Navi to come out because that's what's going to tell us most about next gen consoles.

 

I'm really banging my head against a wall here.

I'm not trying to prove the tastes of gamers. I'm trying to ascertain whether the assertion "PC gamers are more interested in framerate" is in any way, even one little poll, correlated with actual data. I am well away of the limitations of a tiny little poll in a tiny little specialist corner of the graphics enthusiast market, but anything is better than nothing.

My point is if no-one has meaningful data, no-one should be stating these points as fact. I do not state that gamers aren't interested in framerate because my data isn't substantial enough to prove that. However, it does show one negative correlation between PC gamers and framerate.

The great irony here is if the results of that poll were heavily weighted in favour of framerate, people would be more willing to believe in its results...

There is no data so no-one should be making any assumptions about what PC gamers prefer.

That doesn't mean PC gamers prefer framerate. There could be a smaller proportion of gamers who value framerate but are able to push it higher than consoles. The total interest in framerate could be 0.1% of PC gaming, competitive gamers, while the rest may prefer < 30fps graphics whoring, for all we know.

As no-one has any data that's meaningful, no-one should be basing logical arguments based on a completely guessed behaviour of a population.

 

Why should it shock anyone? It was publicly announced in March 2018, it didn't just pop into existence out from nothing. All hardware vendors and most likely all AAA-class studios obviously know about these way before the public does.

 

Did MS say that the Halo engine demo was any form of Xbox target? If not it could easily be the halo PC engine target.

Is there going to be a Forza motorsport in 2019, or was it just a rumour I remember hearing that they was taking more time on 8?

So even if you believe you spotted some RT in the halo demo (not saying your wrong), I personally don't see that as a conformation/reason to expect it in the next console. As we know it's going to pc also.

I do believe MS engines will have RT built in, just not convinced it means it will be used on console.

 

The main problem is that existing solutions are not easily to adopt to new technology. The same as with Vulkan and DX12 where only a few developers rebuild the engine completely. This rebuilding applies in particular to titles that have purpose built engines. Other engines such as Unreal Engine 4 or Frostbite cover many users and despite the good employees it is not easy to rebuild the whole game engine. The existing market is too big with existing toolchains etc. These engines earn their money with stability for the developers.

On the other hand some smaller studios, SEED or the new studio of ex-SEED employees are building a completely new technology stack. Everything runs on a long term basis until the features are optimally used. Therefore it is even more important to bring it into the market early.

 

Actually it looks "night and day" better. Not in screenshots but when you play the game. Rotterdam has animated birds which can "fly" over the channel in the middle of the map. The SSR is moving synchronize with the position of the bird. It is like the FALD zone tests of TVs where you can see the dedicated zones moving with the white square.

I think what is missing from the discussion is the design aspect of the levels. DICE uses SSR in such a excessive way that you cant miss all the problems. With DXR their vision is now realisable.

 

I’m responding to the idea that it may be too early for DXR to influence consoles. As you’ve indicated by the time consoles are ready to go out it would have been 3+ years.

I don’t think that’s too early at all

 

I'm pretty sure RTRT will be a thing, by then. So yes, maybe Sony/Microsoft will be compelled to include some kind of hardware RT tech in their consoles OR do a special focus on power and how good their games will look despite not having RT tech.

 

You’re not wrong here. In my post of assumptions these are the effective counter points.

For me this argument lies on pro-RT people to convince everyone why it should move forward on console, because the opposing argument, let’s be real, there are a million and 1 reasons not to put in RT into consoles.

My answer has been straight forward though, by moving RT into the console space, ideally both, we can leverage the technology to create all sorts of effects and complexities at lower costs to developers. Bring up the graphical fidelity across the board without needing to rely on first party titles to tow the line. Be able to see all sorts of beautiful graphics for all types of games, and not just TPS adventure titles and racing games.

I think for a platform perspective looking at this technology; how effective it is to be the top dog for something as simple as winning DF enough through both resolution and some better AF and draw distance. i think the impact of having RT would put your entire library ahead of your competitor that doesn’t have it in almost every scenario where RT is used.

Where Sony first party studios spend exorberant time and money to produce their visuals, the problem would be solved using RT for any company with a fraction of the cost. That’s a pretty big equalizer here. Theres only so much talent out there and being able to navigate every single draw back and make every single thing interlock and work together is mighty expensive. We’ve only seen Rockstar be the only 3P company to really attempt it. No one else does honestly.

We’ve only seen reflections so far, AO, GI, and shadows are still coming with other titles releasing this year. And they will get better, much better than the solutions we've seen in the past and much simpler to boot. Even if there wasn’t enough power to do reflections, that doesn’t limit everything else RT acceleration can help with.

I can only think one would feel that they could get a leg up on their competition by having it and take a risk of next gen banking on this over any other emergent technology.

So for me I can only hope that both do by thinking the same way.

 

Yes. I think from a perfect standpoint that yes, rebuilding an engine from scratch is expensive. But we run into those issues today with current rasterization as well. The difference is, you can still bolt on DXR like we saw wth battlefield. But there can be some unforeseen bugs that perhaps are how you solve things in rasterization but could cause issues in your ray tracing path.

But the win here is that after some work you can still come out with RT effects on an engine that you may not normally be able to.

 

I definitely think the discussion needs to be put on hold until we see some real-time lighting performance. If GI can be 'solved' effectively (quality, res and framerates) with RT hardware, I think it's valuable. One of the other considerations, using RT for non-graphics work, seems implausible now though for performance reasons. RT will be fully utilised just producing the visuals without spare runtime for tracing other elements, which can be left to the CPU (pending some paradigm shift in how game world's are represented and visualised, maybe wherein tracing can be used for world building that includes AI and sound traces as part of the model within the ray casting results, if such a thing is even possible).