GPU Ray Tracing Performance Comparisons [2021-2022]

[iroboto]

It's understandable that some regular run of the mill dude will often not give a flying fuck about graphical progress and enhancements, but it's not understandable when the tech elite and graphics enthusiasts do that exact same thing in a forum dedicated to graphics and techs.

But I guess that's what happens when you affiliate yourself to a certain vendor, you convince yourself that RT brings nothing to the table so that you can buy current/previous gen AMD hardware and not feel bad about missing out.

There may be some miscommunication happening here which may be the stem of frustration. Every story has a beginning, and the tale of RT has really just begun. Not everyone is interested in reading a book based off the first few pages, some people need to know it's going to be great and hear amazing great things about it before they get onboard. There's nothing wrong with being an early adopter, and equally there's nothing wrong with people hanging back either.

But I what I would agree with is that, it's important for groups to recognize that Ray Tracing needed to start somewhere and that starting point was always going to be tame not because of the hardware, but because of how long software would take to catch up. It's still many years out, so I understand why some would not be interested in being an early adopter, but equally I find it unfair to dismiss it before it's had a chance to blossom either.

 

[Svensk Viking]

No, it's because he has ruined the reputation of a once venerable site: TechSpot.


People, not tech reviewers, those represent products and what they are capable of.

There is a standard called DX12U, with it comes new graphics features calles ray tracing, not testing your new hardware with the new graphics standards represents the most stupid and misleading thing I have seen in tech journalism in decades.

HBU had no problem using early DX12 tests in their reviews despite them not offering anything new to the table, not even performance enhancements (on the contrary, it often made performance worse), but now when DX12 has new graphics features they don't test because it doesn't matter!! That's not tech journalism, that's hypocrisy.

And it bit them in their assess, now they have to go through hoops to justify favoring the featureless RDNA1 GPUs over Turing.

Tech reviewers are not supposed to be representing any product. That is the whole point of getting independent third party reviews. And they are fully free to motivate their choices for what settings they use and do not use, if enough people do not like it they will notice it by losing viewers.

If HBU did constantly present only DX12 results from Nvidia, and never the better DX11 results despite the graphics being exactly the same, that is indeed questionable, there is no doubt about that. Perhaps they decided to focus on DX12 performance because they thought would become more important faster than it did, but yes, questionable indeed.

You also come off as a hypocrite however, seeing that you complain about them testing only the newest API back then because Nvidia was worse, but now all of a sudden you are complaining that they are not testing and proclaiming the greatness of the newest features, for which Nvidia happens to have a big advantage.

If more people were upset they cannot play Minecraft RTX or Quake 2 RTX you would probably see more commotion about it.

 

[techuse]

HUB typically used the highest performing API for a given vendor. I have no idea what he's talking about.

 

[Svensk Viking]

There may be some miscommunication happening here which may be the stem of frustration. Every story has a beginning, and the tale of RT has really just begun. Not everyone is interested in reading a book based off the first few pages, some people need to know it's going to be great and hear amazing great things about it before they get onboard. There's nothing wrong with being an early adopter, and equally there's nothing wrong with people hanging back either.

But I what I would agree with is that, it's important for groups to recognize that Ray Tracing needed to start somewhere and that starting point was always going to be tame not because of the hardware, but because of how long software would take to catch up. It's still many years out, so I understand why some would not be interested in being an early adopter, but equally I find it unfair to dismiss it before it's had a chance to blossom either.

What is annoying me is that I get this impression that there is a genuine hate and spite towards anyone who says they do not see the current importance/benefits of raytracing. This last page has even showed the mindset that no one but qualifid experts of the field should be allowed to make reviews.

I can fully buy that some people prefer raytracing already right now, and that some do not. But until we see it actually being the minimum requirement, be it already next year or even five years from now, I think the pro-raytracing camp should hold their thoughts because all they are doing is creating toxicity on the forum.

 

[Deleted member 2197]

I think the pro-raytracing camp should hold their thoughts because all they are doing is creating toxicity on the forum.

Funny you can't see that the ray tracing demonizers are creating the toxicity because they aren't in a position to enjoy what ray tracing brings to the table. Maybe when the hardware they use is capable their minds will change.

 

[DavidGraham]

But until we see it actually being the minimum requirement, be it already next year or even five years from now

Such a weasel requirement, no graphics features has ever became minimum requirement this fast, this never happened with DX7 or 8 or 9 10 or 11 or 12, and it will never happen with Ray Tracing this quickly.

But I see that opponents of ray tracing have to come up with ridiculous impossible goal posts to justify their non sensical arguments and disdain towards RT.

I think the pro-raytracing camp should hold their thoughts because all they are doing is creating toxicity on the forum.

See the paragraph above, on the contrary, it's the opposite side who is coming up with mind bending arguments to ridicule a graphics technology in a tech forum.

Tech reviewers are not supposed to be representing any product.

Nope, that's just absolute and complete crap, no body gives a damn about a channel like Linus because they never claimed they are expert entrenched tech reviewers, but HBU is coming off the legacy of TechSpot, and as such will be treated accordingly.

If HBU did constantly present only DX12 results from Nvidia, and never the better DX11 results despite the graphics being exactly the same,

They did that all the time. Your ignorance toward their history is curious given your constant defense of them.

HUB typically used the highest performing API for a given vendor

Nope, a quick look at the history of their reviews and you will see what I am talking about, even now they test Borderlands 3 only in DX12 despite DX11 offering higher and stable fps.

 

[iroboto]

What is annoying me is that I get this impression that there is a genuine hate and spite towards anyone who says they do not see the current importance/benefits of raytracing. This last page has even showed the mindset that no one but qualifid experts of the field should be allowed to make reviews.

I can fully buy that some people prefer raytracing already right now, and that some do not. But until we see it actually being the minimum requirement, be it already next year or even five years from now, I think the pro-raytracing camp should hold their thoughts because all they are doing is creating toxicity on the forum.

Communication isn't an easy thing and it can be very difficult to recognize what people are trying to say when not everyone can properly communicate it, or if people are not "really, really" listening to what people are trying to say.

It may not always be that either, it may very well just be a circumstance of perspective. Some people placing a lot more weight on futures, while others placing a lot more weight on 'today'. RT does need a dedicated following to continue to push forward to become a standard, so that it'll happen sooner. Otherwise, without support, they would have to wait until RT becomes cheaper (labour wise) to make RT a standard, and that will happen at the speed of economics.

 

[Svensk Viking]

Funny you can't see that the ray tracing demonizers are creating the toxicity because they aren't in a position to enjoy what ray tracing brings to the table. Maybe when the hardware they use is capable their minds will change.

You can say that all you want, all I have seen here is a few proponents of raytracing getting super upset about a single reviewer publishing his results with raytracing off with the motivation of not being worth the performance loss, and then even more upset reactions towards us "ray tracing demonizers" saying a reviewer has the full right to do his tests according to his own preferences.
Worst case, just ignore the guy and check other reviews, and this discussion would never have continued page after page.

But that you admit that the hardware needs to be capable shows you might have finally gotten a piece of the other side of the argument.

Such a weasel requirement, no graphics features has ever became minimum requirement this fast, this never happened with DX9 or 10 or 11 or 12, and it will never happen with Ray Tracing.

But I see that opponents of ray tracing have to come up with ridiculous impossible goal posts to justify their non sensical arguments and disdain towards RT.

What the "opponents of raytracing" have come up with is that the feature currently is too costly for what it brings to the table. Actually quite similar to how it was back in eg DX11 and the tessellation era, and back then people did not say the DX10 cards now were useless for modern games due to lacking the new features.

Stop making non-sensical arguments like "featureless RDNA1 GPUs over Turing." until it actually prevents people from playing the new games.

See the paragraph above, on the contrary, it's the opposite side who is coming up with mind bending arguments to ridicule a graphics technology in a tech forum.

It is not about ridiculing a graphics technology, it is about questioning the current state of its current usage and beneficialness in a game. A tech forum is indeed a place where you can discuss different technical trade-offs, and not a place where you are required to defend or endorse a specific technology.

Interesting that you choose to look at all negative comments about raytracing as ridiculing the technology however.

Nope, that's just absolute and complete crap, no body gives a damn about a channel like Linus because they never claimed they are expert entrenched tech reviewers, but HBU is coming off the legacy of TechSpot, and as such will be treated accordingly.

Your answer is absolute and complete crap and shows you are not even understanding the sentence. A third party reviewer is not the representative of a product, that would defeat the whole purpose of independent third party reviews.

They did that all the time. Your ignorance toward their history is curious given your constant defense of them

Because compared to you I do not make a list of reviewers I hate and which reviewers I would gift a kidney

I am very open when checking reviews, looking at alot of different sites and alot of comparisons for the various settings. I can skip reading a review if it is clear he has other priorities than me, but I am never making posts on forums telling people to skip it due to it being invalid because we have different opinions of what settings/performance are most important for the game.

 

[DavidGraham]

Stop making non-sensical arguments like "featureless RDNA1 GPUs over Turing." until it actually prevents people from playing the new games.

It actually does, it prevents you from accessing higher quality versions of Metro Exodus, Quake 2, Minecraft, and you are completely locked out from even launching a game called Stay In The Light. Not to mention not being able to access max settings in dozens of games. More to come in the future of course.

Interesting that you choose to look at all negative comments about raytracing as ridiculing the technology however.

Because all you do is exactly that, you can't even acknowledge the effect of RT in the games that use it the best, in addition to posting some impossible goal posts that was never met by any prior technology.

A third party reviewer is not the representative of a product, that would defeat the whole purpose of independent third party reviews.

No actually, a third party reviewer have to do exactly that, test fps, power consumption, heat, noise, latest tech .. etc, this is the standard required from EVERY damned reveiwer, omitting a big fat DX12U from the equation just because he deems so only makes him unprofessional and hypocritical, especially in light of him testing with DX12 back when it was absolutely useless.

Because compared to you I do not make a list of reviewers I hate and which reviewers I would gift a kidney

You should just pay attention before blindly defending a reviewer just because he agrees with your stance against ray tracing.

 

[Svensk Viking]

It actually does, it prevents you from accessing higher quality versions of Metro Exodus, Quake 2, Minecraft, and you are completely locked out from even launching a game called Stay In The Light. Not to mention not being able to access max settings in dozens of games.

None of which is a concern for people today who think raytracing not is worth the performance loss. The problem for people without RT HW will appear when there only is a RT-required version available.

Because all you do is exactly that, you can't even acknowledge the effect of RT in the games that use it the best, in addition to posting some impossible goal posts.

I have said that some people like raytracing already today, and compared to you, I have never said they should not be allowed to talk of how great it is or invalidate their reviews. What I am opposed against is this echo chamber of yours saying the pro-raytracing crowd is the only one allowed to speak and that anyone who disagrees is dead wrong and should be put to silence.

It is your argument that RDNA1 now is useless, and mine is that it is not useless until RT actually has become the minimum requirement. That you actually get upset once again should be food for thought.

No actually, a third party reviewer had to do exactly that, test fps, power consumption, heat, noise, techs, this is the standard required from EVERY damned reveiwer, omitting a big fat DX12U from the equation just because he deems so only makes him unprofessional, especially in light of him testing with DX12 back when it was absolutely useless.

What makes a reviewer truly unprofessional is if he agrees to terms from the company sending him the product to only test it in ways which show their advantages over competing hardware. Which actually, as noted earlier in the thread, Nvidia required HWU to start doing.

Also, a review can focus on anything. It can be gaming performance at 30% TDP, it can even be focusing on Nvidia Linux gaming performance with the Nouveau drivers. You really are hard set that everything needs to be done exactly the way you like it or it is dead wrong and invalid.

You should just pay attention before blindly defending a reviewer just because he agrees with your stance against ray tracing.

I am defending him because he has the right to publish his results without getting vetoed by an Nvidia committee. Not because as if we were in full agreement.

And you should stop letting your bias from your obvious earlier hate history with them blindly slander their reviews. Out of nowhere, the complaints towards the guy turned from being his thoughts of raytracing to being about the history of Techspot.


A forum is a place where you are supposed to discuss and be able to accept reading posts that you do not agree with. If you are only thinking in absolutes a forum is the wrong place to be.

 

[DavidGraham]

None of which is a concern for people today who think raytracing not is worth the performance loss. The problem for people without RT HW will appear when there only is a RT-required version available.

It is a concern, if you are buying a GPU for 400$ or more, only to be locked out of graphics as if you bought some low end GPU. It's actually worse. It's frustrating to anyone.

Please don't claim to speak on behalf of all of the people, this is a tech forum, we evaluate hardware based exactly on those merits.

I have never said they should not be allowed to talk of how great it is

Yes you did:

I think the pro-raytracing camp should hold their thoughts because all they are doing is creating toxicity on the forum.

What makes a reviewer truly unprofessional is if he agrees to terms from the company sending him the product to only test it in ways which show their advantages over competing hardware. Which actually, as noted earlier in the thread, Nvidia required HWU to start doing.

Because this is a god damned industry standard, not just a vendor feature, RT is going into mobile for God's sake, it's in consoles and console games are featuring it in droves.

HUB is going out of it's away to omit testing such an industry standard, twisting his way out of it with unconvincing arguments -despite their previous tendency to test with latest APIs-, only to back pedal on it later on.

I can't believe you still don't get all of that, please if you don't understand the level of hypocrisy here then get out of the conversation. You are not adding any more to it, and you are failing miserably in keeping up with the history of said YouTube channel or their ever changing stance on the matter.

 

[Svensk Viking]

It is a concern, if you are buying a GPU for 400$ or more, only to be locked out of graphics as if you bought some low end GPU. It's actually worse. It's frustrating to anyone.

Please don't claim to speak on behalf of all of the people, this is a tech forum, we evaluate hardware based exactly on those merits.

Whether it is a concern or not depends what each individual buyer of said GPU thinks. You say not to speak on behalf of all the people, so live as you preach.

It is a tech forum and a forum is where different opinions are voiced.

Yes you did:

In the context of when raytracing really is going to be a requirement to be playing games, yes. Are you trying your hardest to interpret everything as hostile to your love for raytracing? I still actually want the benefit of doubt here so I hope it is the language barrier.

Because this is a god damned industry standard, not just a vendor feature, RT is going into mobile for God's sake, it's in consoles and console games are featuring it in droves.

HUB is going out of it's away to omit testing such an industry standard, twisting his way out of it with unconvincing arguments -despite their previous tendency to test with latest APIs-, only to back pedal on it later on.

I can't believe you still don't get all of that, please if you don't understand the level of hypocrisy here then get out of the conversation. You are not adding any more to it, and you are failing miserably in keeping up with the history of said YouTube channel or their ever changing stance on the matter.

It is an industry standard, and just like for the standards in the past, reviewers have had the choice if testing only DX8 or DX9 or DX11, or not testing DirectX at all because they decided only to test the OpenGL performance, they can choose to test with tessellation on or off.
His argument is apparently unconvincing to you but that does not invalidate the performance figures for the settings he are using.

I understand that you are greatly upset that I disturbed your echo chamber, but you should take that as a sign of what forums really are for. You seem to take everything you do not agree with personally and that is not good for your health long term.

 

[neckthrough]

Folks, this is getting ugly and pointless. Let's all take a deep breath.

The only real disagreement here is where exactly we are on the timeline of adoption of RT as the standard light transport mechanism. There's nothing to be gained from arguing back and forth.

Instead there are tons of other interesting topics for discussion, and serious technical challenges that still need to be overcome. Denoising. The cost of maintaining multiple lighting pipelines. Hybrid approaches like DDGI/RTGI as graceful transition paths (this one particularly interesting). Or just simply strengths/deficiencies in specific implementations in games. That's the stuff we all want to be talking about.

Some suggestions to keep ourselves calm:

• Proponents have to stop being triggered by the occasional negative sentiment from a skeptic. Ignore and move on.
• On the flip side, the next time we discuss a game with a poor implementation, there's no need for skeptics to generalize that as an argument for "yeah, RT still sucks". Comment on the specific game and its problems.
• No hardware vendor/platform bashing. Yes, there's a vendor disparity at the moment. But we've still seen superb implementations on very meagre RT hardware. We'll see what the future brings.
• No bad faith arguments. We all know when we're making one. Let's exercise some self restraint.
• No meta-commentary on tech-tubers. Calling out bias is important, but not helpful within the context of this thread.

Bottom line is - stay technical, stay objective and stick to the topic at hand.

Sorry I know this was preachy but FWIW I've been guilty of some of these points myself.

 

[iroboto]

Folks, this is getting ugly and pointless. Let's all take a deep breath.

The only real disagreement here is where exactly we are on the timeline of adoption of RT as the standard light transport mechanism. There's nothing to be gained from arguing back and forth.

Instead there are tons of other interesting topics for discussion, and serious technical challenges that still need to be overcome. Denoising. The cost of maintaining multiple lighting pipelines. Hybrid approaches like DDGI/RTGI as graceful transition paths (this one particularly interesting). Or just simply strengths/deficiencies in specific implementations in games. That's the stuff we all want to be talking about.

Some suggestions to keep ourselves calm:

• Proponents have to stop being triggered by the occasional negative sentiment from a skeptic. Ignore and move on.
• On the flip side, the next time we discuss a game with a poor implementation, there's no need for skeptics to generalize that as an argument for "yeah, RT still sucks". Comment on the specific game and its problems.
• No hardware vendor/platform bashing. Yes, there's a vendor disparity at the moment. But we've still seen superb implementations on very meagre RT hardware. We'll see what the future brings.
• No bad faith arguments. We all know when we're making one. Let's exercise some self restraint.
• No meta-commentary on tech-tubers. Calling out bias is important, but not helpful within the context of this thread.

Bottom line is - stay technical, stay objective and stick to the topic at hand.

Sorry I know this was preachy but FWIW I've been guilty of some of these points myself.

It's a good time for all parties to just take a break from this thread and just let time do its work. You may find with some breathing space you may look at things differently etc. or come to a common ground that this isn't worth continuing.

 

[DavidGraham]

Far Cry 6 has visibly lower resolution RT reflections on AMD GPUs compared to NVIDIA GPUs.

https://imgur.com/IKnZBJR

https://imgur.com/uGSaZ5P

https://imgur.com/a/ukGxJLK

 

[milk]

Shadows, fog, particles, volumetric lighting, shaders, tessellation, ambient occlusion, reflections, DOF, AA are all refinements. They all add to the final result.

The irony is that one of the main reasons people don't fully appreciate RT yet is that games aren't designed with RT as a baseline. If you look at a scene with one or two shadow casting lights it's going to be hard to see where RT helps because advanced shadow mapping (and caching) techniques look really good. But games don't include scenes with many dynamic shadow casting lights because shadow mapping would be laughably slow. So when people say RT shadows are blah it's understandable because they have no frame of reference for what real-time shadows "should" look like in more complex scenes.

Well, more shadow casting lights also means more rays in an RT solution. That or a very sparse stochastic sampling of the lights, which would probably look very blurry and possibly worse than shadowmaps.

 

[DegustatoR]

Some nice "image refinement" happening here:

https://i1.lensdump.com/i/gQxpz3.jpg
https://i2.lensdump.com/i/gQx7w0.jpg

That's from what is more or less a very mediocre RT implementation in RE8.

HUB typically used the highest performing API for a given vendor. I have no idea what he's talking about.

Literally never seen them doing that.
They use DX12/VK when possible, even when said options are lowering performance on all GPUs they are testing.
The only website which I'm aware of which is doing what you're saying is computerbase.de.

I must admit I did find Hardware Unboxed conclusion on the RX6600 funny

Well, he did managed to say that "6600 is going to be the best option if you absolutely need to buy a new GPU".
And I honestly fail to see how exactly he has arrived to that conclusion.

 

[Davros]

I must admit I did find Hardware Unboxed conclusion on the RX6600 funny

Expensive 5600xt but with really sh*t raytracing

 

[trinibwoy]

Well, more shadow casting lights also means more rays in an RT solution. That or a very sparse stochastic sampling of the lights, which would probably look very blurry and possibly worse than shadowmaps.

It wouldn’t mean more rays as each pixel would only sample a few lights to determine if it’s in shadow. With importance sampling the total number of rays should be pretty constant as you increase the number of lights.

 

[milk]

It wouldn’t mean more rays as each pixel would only sample a few lights to determine if it’s in shadow. With importance sampling the total number of rays should be pretty constant as you increase the number of lights.

I understand, but Nvidia has shown some demos with importance sampling for many lights, and the result was that the shadows became so blurry they weren't as visually impressive one would think.