[FREE] Pure RayTracing and Non-RT benchmarks

[miguel1900]

Large part of the scene is still done with rasterization and other more common methods. There are certain effects or buffers (or parts of them as with your example of Skybox lighting getting raytraced at higher settings) that are raytraced (as a general term, whether the actual algorithm/method used is pathtracing, some hybrid and/or the nice postprocessing like inevitable denoising) that are then composited together. Reflections, shadows, GI/AO are the main ones with UE as far as I remember.

UE does have some type of full pathtracing renderer but to the best of my knowledge that is essentially for reference renders, not real-time usage. From what I have seen or heard from its performance, this is absolutely definitely not what your benchmark uses

Edit:
UE RT overview in their own documentation:

Hardware Ray Tracing Tips and Tricks in Unreal Engine | Unreal Engine 5.4 Documentation | Epic Developer Community

A collection of topics to aid with development of projects using hardware ray tracing features in Unreal Engine.

docs.unrealengine.com

Translucency is the one I forgot. GI and AO are separate as well.

I only have the doubt about which part is done with rasterization and how to check it. But you are totally right: there are denoisers, of course, and yes, actual Path Tracing it isn't for real time. RT is something like a "light" PT. Well, in fact, certain very little effects are slightly different between RT and PT, thinking about it should make the calculations in a slightly different way, that was what you meant? Well, anyway yes, it's not fully path traced, hehe, but fully ray traced.

Regards!

hmmm well i think a 3080 will be an upgrade even with the 10GB of memory

Of course! haha. Thanks for sharing

 

[londiste]

While I am at it, 2160p results as well, although these are pretty slideshow on my hardware

2160p Common: 13.09 FPS
2160p High: 9.04 FPS
2160p Pro: 7.31 FPS

Spoiler

VRAM usage peaks at 7.5GB with Pro settings and depends more on settings than resolution.
Common settings use 500-600MB less but still peak at 7.5GB at one point.

These two scenes or moments are the best eyecandy IMO:

Spoiler

I only have the doubt about which part is done with rasterization and how to check it. But you are totally right: there are denoisers, of course, and yes, actual Path Tracing it isn't for real time. RT is something like a "light" PT. Well, in fact, certain very little effects are slightly different between RT and PT, thinking about it should make the calculations in a slightly different way, that was what you meant? Well, anyway yes, it's not fully path traced, hehe, but fully ray traced.

Simple - anything that is not reflections, shadows or lighting is rasterized. And no, it is not fully raytraced. Shadows are (or might be, depending on settings), same for reflections and GI/AO taking them separately but not the scene as a whole

Raytracing works fine as a general term in this context. PT and RT are pretty interchangeable when we talk about hybrid rendering and what is rasterized or raytraced. It is a technical question, which method is used by UE you can probably ask or look it up but I would assume it is more likely PT.

 

[miguel1900]

While I am at it, 2160p results as well, although these are pretty slideshow on my hardware

2160p Common: 13.09 FPS
2160p High: 9.04 FPS
2160p Pro: 7.31 FPS

View attachment 170287 View attachment 170288 View attachment 170289


VRAM usage peaks at 7.5GB with Pro settings and depends more on settings than resolution.
Common settings use 500-600MB less but still peak at 7.5GB at one point.

These two scenes or moments are the best eyecandy IMO:

View attachment 170290 View attachment 170291


Simple - anything that is not reflections, shadows or lighting is rasterized. And no, it is not fully raytraced. Shadows are (or might be, depending on settings), same for reflections and GI/AO taking them separately but not the scene as a whole

Raytracing works fine as a general term in this context. PT and RT are pretty interchangeable when we talk about hybrid rendering and what is rasterized or raytraced. It is a technical question, which method is used by UE you can probably ask or look it up but I would assume it is more likely PT.

Thank you!

Didn't thought about things not being reflections, shadows, lightings... could be also rasterized. Anyway, nothing to add, hehe, all you said is totally right!

Oh, about VRAM, I also tested on a GTX 1060 6GB and it worked fine. I think the more you have, the more it gets.

 

[londiste]

More details about what settings exactly are changed for High-Pro would be nice as well.

Ha, this one is on me not reading the link correctly.

                         Common  High    Pro
RT Reflections           ON      ON      ON
RT Shadows               ON      ON      ON
RT Ambient Occlusion     OFF     ON      ON
RT Global Illumination   OFF     ON      ON
RT Sky                   OFF     OFF     ON

 

[miguel1900]

Ha, this one is on me not reading the link correctly.

                         Common  High    Pro
RT Reflections           ON      ON      ON
RT Shadows               ON      ON      ON
RT Ambient Occlusion     OFF     ON      ON
RT Global Illumination   OFF     ON      ON
RT Sky                   OFF     OFF     ON

And me, not reading that comment. Great schema! Adding it to the first post, thank you!

EDIT: Ouh, it seems I can't edit the first post anymore.

 

[stealth83]

Msi SeaHawk X 2080ti stock

 

[scope54]

Ryzen 2600x and Radeon 7

GPU-z Reports max Dedicated memory used during the 1440 pro run was about 4GB.

 

[miguel1900]

Msi SeaHawk X 2080ti stock

View attachment 170298

View attachment 170299

Ryzen 2600x and Radeon 7

GPU-z Reports max Dedicated memory used during the 1440 pro run was about 4GB.

View attachment 170300
View attachment 170301
View attachment 170305

Thank you for sharing!

Unfortunately, @scope54 , AMD cards are currently non capables of Ray Tracing so, probably, if you compare your sequences with the sequence shown in the video of the first post, you should notice clear differences.

But there will be a non-RT benchmark version soon!!

 

[wolf]

3700X and RTX3080

 

[miguel1900]

Thank you @wolf ! first 3080 over here. Added to the scores table!

 

[birdie]

@miguel1900

I don't think it will take too much of your development time if you implemented some sort of proper summary window, like e.g. in Unigine benchmarks:

Please, do.

Here's my GeForce 1660 Ti FHD Common Settings results (stock) - 13.27 fps along with Ryzen 3700X (stock).


Not that they are interesting to anyone here considering that I have what could be called an entry level GPU.

 

[miguel1900]

@miguel1900

I don't think it will take too much of your development time if you implemented some sort of proper summary window, like e.g. in Unigine benchmarks:

Please, do.

Here's my GeForce 1660 Ti FHD Common Settings results (stock) - 13.27 fps along with Ryzen 3700X (stock).


Not that they are interesting to anyone here considering that I have what could be called an entry level GPU.

Hi @birdie ,

Thanks for sharing! Of course, every card result it's important for every kind of player. It's already added to the official list.

About that screen, probably I will end adding it. The problem is that I didn't want to "cover" the screen with that message after the test, to allow the user to play. So I decided to add the final info too when you exit from the benchmark, going back to menu, in a descreet line at the bottom of the menu, but I'm not sure if some of you haven't seen it, or if you want a little more info, like min and max FPS (which was already asked by another user), or if you simply want an independent summary window, so I'm open to suggestions!

 

[miguel1900]

Hi guys!

I'm working on some great things.

The next one will be the same benchmark but without RayTracing, to be a benchmark for any kind of user (well, DX12 required), released independently.

But next... I will unify the two benchmarks, RT and non-RT, in the same one, simply checking the option, and I'm adding some great additions to the 'menu' screen, which will be a menu + statistics screen now, as never seen before in any other benchmark on the market! (I think! So, please, if you want, you can post here a screenshot of the most 'analytical' benchmark screen you have ever seen).

More details about its new options coming soon.

Best regards!

 

[miguel1900]

Hi all!

I'm glad to announce that this RT benchmark has been updated to v1.5. I have heard all your comments, fixed some minor things and added new options (some still in development, for the next release! That one will be awesome!).

I have also launched a version without Ray Tracing capabilities, so any kind of user with a DirectX12 compatible graphic card, will be able to execute it and compare his measurements with others. I have been suggested to keep those non-RT results in this same thread too, but if we notice it becomes a little chaotic, I will ask our moderators again about opening an independent new thread.

Free Non-RT download: https://marvizer.itch.io/pure-non-rt-benchmark

(The upcoming v2.0 release will unify these two benchmarks in only one, being able to switch between then selecting one simple option. In fact that option already exists, but it's currently disabled, as it's still in development. But it works like a reminder).

I hope you like it!

PS: Oh, I can't edit the first post anymore, right?

 

[k1llsh0t16]

I have a RTX 2080 Super but the benchmark is saying my system can't do ray tracing. Any thoughts? RTX is definitely working in games and other benchmarks.

 

[miguel1900]

I have a RTX 2080 Super but the benchmark is saying my system can't do ray tracing. Any thoughts? RTX is definitely working in games and other benchmarks.

Hi @k1llsh0t16 ,

Thank you very much for your feedback. Please, could you make some tests to check it and try to solve? My bench is checking, in a pragmatical way, if your PC is actually executing RT, during the firsts 0.2 seconds. On my old PC, with a GTX 1060 it was more than enough to successfully check it. So, maybe, your PC is really not running RT, or it has a huge bottleneck to check it before the firsts 0.2 seconds, or an even more strange case, but I will be ready to deep into the problem.

Please, could you try starting testing this mini tool?

File on MEGA

mega.nz

And this Nvidia's one? (I remember I tried it when my PC wasn't RT compatible, and it shown me a warning):

https://images.nvidia.com/geforce-com/international/reflections-demo/Reflections_Demo.zip

If nothing, I would want you to try another thing, please.

Thank you and best regards!

 

[miguel1900]

I have now published the mini-tool to check if a computer is actually executing the Ray Tracing technology, in an easy and ultrafast way: https://marvizer.itch.io/ray-tracing...bility-checker

 

[Lord KUKO]

Hey. I have win 10 x64 and RTX 3090 but please why i have this? https://prnt.sc/vs22kk

 

[miguel1900]

Hey. I have win 10 x64 and RTX 3090 but please why i have this? https://prnt.sc/vs22kk

Already talked about this with Lord KUKO trough private messages! Will help to check it for a future fix. It's an unknown error for some GPUs, which can check if RT is enabled in time, for some reason.


PD: some RX 6800 XT results added to the score tables!! They are almost like a 2080 Ti

 

[Hugis]

1080ti here the app returns RT capable and the test returns non RT

 

[Athlonite]

looks great runs ok on my system ok aswell Sapphire Pulse RX5700 8GB @1750cclk/1700mclk and R7 3700X @ 4325MHz

 

[Zyll Goliat]

Here for your Data-Base NoN RT result from my R9 Fury

 

[miguel1900]

Thank you all!

1080ti here the app returns RT capable and the test returns non RT


View attachment 177717

Hi Hugis,

You are right. I have noticed some few users having this problem. For some reason, your computers can't check if RT is enabled or disabled in the given time. The 'checker tool' have been designed with a improved 'more flexible' method, which will be added in the next release of the benchmark, coming very soon. Sorry for the inconvenience!

 

[Caring1]

Compatibility checker wouldn't run for me.
First it was telling me I needed to install Visuals C++ 2015, although it is already installed, then it was telling me after I redownloaded that, that I need DX to run it, even though I have DX12.

 

[fullinfusion]

@miguel1900 Nice work man! I just tested the non RT demo on a Vega card and would you believe it your benchmark auto kicked in my monitors HDR and it looked GREAT!!

Great job!