GPU Ray Tracing Performance Comparisons [2021-2022]

it has not been announced yet - but I imagine they may use the Surfel GI based on the presentation.
Hmm, I consider that to be an advanced ray tracing algorithm. Similar to how a grid of irradiance probes can be used in an advanced ray tracing algorithm. In both cases rays are used to build/use the acceleration structure.
 
Hmm, I consider that to be an advanced ray tracing algorithm. Similar to how a grid of irradiance probes can be used in an advanced ray tracing algorithm. In both cases rays are used to build/use the acceleration structure.

It definitely is but there’s been no indication so far that BF6 is using it.
 
Of course not, they should test a mixture of games like they do, but activate RT in those games that support it, as a part of running those games at Ultra/Max settings.

In any fair review right now, half of the tested games will include RT, games like: Metro Exodus, Deathloop, F1 2021, Dirt 5, Call Of Duty, Resident Evil, Cyberpunk, Fortnite, Crysis Remastered, Doom Eternal, Wolfenstein Young Blood, Battlefield V, Control, Watch Dogs Legion, Minecraft, Far Cry 6 .. etc, even Gears 5 should be tested with it's software ray traced Global Illumination, World of Tanks should also be tested with it's software ray traced shadows.

Reviewers should just test those games with RT enabled by default.
I still think these should be separated. While many games have RT, I'd still bet the large majority of people playing them are turning RT off. In most cases, it's still generally quite bad value in terms of visuals vs performance. And because the performance costs are generally quite extreme, we dont have GPU's that have performance to spare to where somebody could just say 'why not?', as can frequently be the case with standard Ultra settings. Turning on ray tracing will usually always mean you're noticeably compromising other areas that PC gamers tend to not like to compromise on(resolution/performance).

So I think the threshold for benchmarking with RT as standard shouldn't be how many games have RT, but how many people are actually using RT when available.

From the footage we have it doesn't look any different than previous BF games. I assume we will see it when Dice drops the old consoles.
From what I've seen, it actually seems to look worse than BF1/BFV? Hard to say so far, but unless they've got very significant graphics updates that they've held back, it does not seem like it's been a focus for them. I think they may have even dialed things back with lighting in order to improve visibility issues that had become a problem in the last couple titles. The game looks flatter as a result, though. A bit more like BF4 than BF1/V(or even Battlefront). We'll see, I guess.
 
I still think these should be separated. While many games have RT, I'd still bet the large majority of people playing them are turning RT off. In most cases, it's still generally quite bad value in terms of visuals vs performance. And because the performance costs are generally quite extreme, we dont have GPU's that have performance to spare to where somebody could just say 'why not?', as can frequently be the case with standard Ultra settings. Turning on ray tracing will usually always mean you're noticeably compromising other areas that PC gamers tend to not like to compromise on(resolution/performance).

So I think the threshold for benchmarking with RT as standard shouldn't be how many games have RT, but how many people are actually using RT when available.


From what I've seen, it actually seems to look worse than BF1/BFV? Hard to say so far, but unless they've got very significant graphics updates that they've held back, it does not seem like it's been a focus for them. I think they may have even dialed things back with lighting in order to improve visibility issues that had become a problem in the last couple titles. The game looks flatter as a result, though. A bit more like BF4 than BF1/V(or even Battlefront). We'll see, I guess.
It does look worse but not in terms of lighting IMO. The terrain tessellation seems simplified as does overall detail levels and LOD. The baked GI looks the same to my eyes.
 
I still think these should be separated. While many games have RT, I'd still bet the large majority of people playing them are turning RT off.
Many gamers play with settings below those which are used for GPU testing. Should we test all settings now?
The purpose of GPU benchmark is to show relative GPU performance, not to guess how people are playing the games used for testing.
RT is already an essential part of a GPU performance. Testing a GPU in a game which have RT with RT being off is misleading and gives results which create a skewed picture for future performance of these GPUs.
That being said there still should be games without RT and these should be used for testing to showcase the difference between RT and non-RT performance.
It is rather similar to how all previous big rendering updates were added into the benchmarks. Hell, people still use CSGO for benchmarking and that's okay if there are modern titles present in benchmarking suite at the same time.
 
Many gamers play with settings below those which are used for GPU testing. Should we test all settings now?
I very specifically went over my reasoning for this already.

People using Ultra settings is very common and there are plenty of GPU's that exist that allow people to just 'turn everything on Max' and then play and have a great experience in many titles.

Ray tracing is quite different. The costs tend to be a lot more extreme and because they're on top of other existing settings, there is rarely ever the overhead to 'just turn ray tracing on' because why not, ya know? You will usually require making some other notable sacrifice to use it.

I'm surprised people dont agree with this. Maybe it's cuz a super tech-focused forum like this is obviously a bit more biased towards the merits of ray tracing than other places, but I dont think this should influence the behavior of benchmarking sites, either. Y'all seem to be asking for these places to cater to our specific, niche whims when that doesn't really make sense from their perspective.

And again, I'm not saying to not test RT performance, I just think such benchmarks should be made separate until it becomes more of the norm. It makes no sense at all to make RT performance 'the standard' when it's still a niche use case. I get the feeling many really want to push this idea because it would strongly favor a certain company in benchmarking, though...
 
I'm surprised people dont agree with this.

The overwhelming majority of people agree with this.
Every single time I see a poll on more mainstream sites like wccftech (like them or not, their audience is huge nowadays and they're now getting exclusive interviews from big players) that asks about what people want from next gen hardware, ray tracing gets pretty much overlooked.

Take their latest poll on what people expect from Arc Alchemist. Better raytracing performance is only desired by 5% of the >700 voters. Top choice is performance per dollar, followed by availability. Even power efficiency gets almost twice the votes of raytracing performance.

IIRC some months ago I saw some stats showing that most RTX3090 users preferred to play with RT off.


There's also the fact that a bunch of developers aren't seeing hardware accelerated raytracing as a must have either.
Ubisoft isn't bringing raytracing to Far Cry 6 on the new-gen consoles, where they preferred to target 4K60 over having raytraced shadows and reflections.
If they thought it was a game changer for visuals, they'd get a lower-resolution mode with it on.
 
Last edited by a moderator:
The overwhelming majority of people agree with this.
Every single time I see a poll on more mainstream sites like wccftech (like them or not, their audience is huge nowadays and they're now getting exclusive interviews from big players) that asks about what people want from next gen hardware, ray tracing gets pretty much overlooked.

Take their latest poll on what people expect from Arc Alchemist. Better raytracing performance is only desired by 5% of the >700 voters. Top choice is performance per dollar, followed by availability. Even power efficiency gets almost twice the votes of raytracing performance.

IIRC some months ago I saw some stats showing that most RTX3090 users preferred to play with RT off.


There's also the fact that a bunch of developers aren't seeing hardware accelerated raytracing as a must have either.
Ubisoft isn't bringing raytracing to Far Cry 6 on the new-gen consoles, where they preferred to target 4K60 over having raytraced shadows and reflections.
If they thought it was a game changer for visuals, they'd get a lower-resolution mode with it on.
As long as RT remains mostly shadows/reflections, I don't expect this mindset to change anytime soon.
 
Take their latest poll on what people expect from Arc Alchemist. Better raytracing performance is only desired by 5% of the >700 voters. Top choice is performance per dollar, followed by availability. Even power efficiency gets almost twice the votes of raytracing performance.

That's not saying much. The basics (price, availability, power) will always trump any one graphics feature. Nobody in their right mind will claim RT performance is more important than price. A more useful poll would be RT vs relevant tradeoffs (higher resolution, 60 vs 120 fps etc).

If they thought it was a game changer for visuals, they'd get a lower-resolution mode with it on.

Or maybe they targeted 60 fps because Far Cry 6 is a first person shooter?
 
That's not saying much. The basics (price, availability, power) will always trump any one graphics feature. Nobody in their right mind will claim RT performance is more important than price. A more useful poll would be RT vs relevant tradeoffs (higher resolution, 60 vs 120 fps etc).
There's actually an option called "Better IQ & Performance With Upscaling Technologies (XeSS)" that got 30% more votes than raytracing performance.
And it's not like people could only choose one over the other. You can choose 3 options in that poll. RT performance is just not a very sought-after feature.


Or maybe they targeted 60 fps because Far Cry 6 is a first person shooter?
I meant the part where they chose to have native 4K resolution over using raytraced reflections and shadows on the Series X.


In a world where temporal upsampling or even just FSR (which is present in the PC version BTW, along with DXR shadows+reflections) is almost always preferred over native 4K resolution, it's quite telling that Ubisoft Toronto's art directors are choosing that instead of raytraced effects.


People's mindset won't change until we've had RT for a while and they go back and play older games. Most people are unable to appreciate things until they're taken away.
It could be that RT will stay relatively dormant for a while (e.g. using it only for hybrid ray traced reflections and little else) until performance picks up, with studios preferring to use cheaper but good enough solutions for global illumination like Epic on UE5, Activision on BF6 and Bluepoint on Demon's Souls.

My guess is unless there's some kind of breakthrough for RT performance on RDNA2 consoles, like idk finding ways to implement some aggressive LOD settings, real-time raytracing isn't going anywhere fast.


We have an example of a GPU feature that actually went away for a couple of generations, and only came back when performance picked up. Supersampling antialiasing was present on DX7 GPUs, went away on DX8, 9 and 10 as games focused more on the less accurate but much more efficient MSAA, and only came back for DX11 GPUs (i.e. when PC GPUs largely surpassed the PS360 in performance).
 
People's mindset won't change until we've had RT for a while and they go back and play older games. Most people are unable to appreciate things until they're taken away.
People have already gone back to non RT shadows and reflections as most games don't support them. They still aren't the areas of visuals most people think need to be focused on for improvement. Those two effects just aren't good choices for such heavy performance trade offs for many people.
 
People have already gone back to non RT shadows and reflections as most games don't support them. They still aren't the areas of visuals most people think need to be focused on for improvement. Those two effects just aren't good choices for such heavy performance trade offs for many people.

Most games aren’t using RT so we haven’t gone back to anything yet. RT shadows are nice but there are shadow mapping and screen space shadowing techniques that already look very good.

Reflections are another story. I’m playing Division 2 a lot these days and there are plenty reflective surfaces using SSR and it’s super obvious when reflections pop in and out of view when the camera rotates. Once you start noticing that stuff it’s hard to unsee.

What I was actually referring to though is that once total lighting solutions like Lumen / DDGI / GIBS take hold the overall look of games will change. Imagine turning off shadows in a game today. Same kinda thing.
 
Most games aren’t using RT so we haven’t gone back to anything yet. RT shadows are nice but there are shadow mapping and screen space shadowing techniques that already look very good.

Reflections are another story. I’m playing Division 2 a lot these days and there are plenty reflective surfaces using SSR and it’s super obvious when reflections pop in and out of view when the camera rotates. Once you start noticing that stuff it’s hard to unsee.

What I was actually referring to though is that once total lighting solutions like Lumen / DDGI / GIBS take hold the overall look of games will change. Imagine turning off shadows in a game today. Same kinda thing.
I agree with your final paragraph. Once RT starts being used for lighting people will be a lot more interested.
 
Of course not, they should test a mixture of games like they do, but activate RT in those games that support it, as a part of running those games at Ultra/Max settings.

In any fair review right now, half of the tested games will include RT, games like: Metro Exodus, Deathloop, F1 2021, Dirt 5, Call Of Duty, Resident Evil, Cyberpunk, Fortnite, Crysis Remastered, Doom Eternal, Wolfenstein Young Blood, Battlefield V, Control, Watch Dogs Legion, Minecraft, Far Cry 6 .. etc, even Gears 5 should be tested with it's software ray traced Global Illumination, World of Tanks should also be tested with it's software ray traced shadows.

Reviewers should just test those games with RT enabled by default.

Yes exactly.


Why though..? Just so the public knows how badly each game plays, using RT..?

Makes no sense, because nobody uses RT... unless you have a $2k card and are playing single player game. Nobody buys a high-end video card to get less performance... they buy a new video card for faster frames...!

FWIW: There is not a single player who plays Warzone with RT on. It is only there (listed) so that NVidia can market it. Even though it has no viability.
 
Back
Top