GPU Ray Tracing Performance Comparisons [2021-2022]

Last edited:
It is not fair at all. Crytek and Saber Interactive are using a hybrid RT approach, so on Turing and Ampere RT gets accelerated unlike RNDA2. Crytek claims it improves performance around 15 to 20%.

But i think it shows how bad RDNA2 as an architecture is. When it comes to compute performance it is barely faster than Pascal on a transistor level. A 3060 can deliver ~60 FPS with maximum RT, a "card designed for 1080p" like the 6600XT is falling from the cliff. So many transistors are spent on something like a L3 cache to hide the fact that RDNA2 is not effcient enough. The best transistor is the one which is doing calculations instead of holding informations.
 
It is not fair at all. Crytek and Saber Interactive are using a hybrid RT approach, so on Turing and Ampere RT gets accelerated unlike RNDA2. Crytek claims it improves performance around 15 to 20%.
Their RT Vulkan interop should work just as fine on RDNA2 cards.
The results there are more interesting for cards without RT h/w though, with 1080 being ~50% faster than Vega 64.
 
What the hell?! Now I have to check this out, holy kaboodle. I never even thought to do a test like this as settings between vendors, by standard, are always the same!!!
Make sure you restart the game between tests after changing graphics settings. Apparently there's a game bug that means changes to settings don't always get applied without a restart.
 
Crytek Crysis Remastered Trilogy controversial review guidelines leaked - VideoCardz.com
Game developers have every right to choose who gets to show the game first and this is indeed often the case with other games. What game developers or any other company providing a product (physical or virtual) for that matter should not do, however, is to tell reviewers what they should, or not say, in their review. The guidelines are only guidelines, and anything other than the reviewer had agreed to beforehand, such as publishing date, should be ignored.

Joker revealed the summary from the review guidelines for Crysis Remastered Trilogy, the latest installment in the series offering better graphics, new graphics features such as ray tracing for the PC version or DLSS for NVIDIA RTX powered systems. But Crytek clearly does not want reviewers to ‘oversell’ graphics, which is a bizarre ‘request’ for a game that has been remastered.

Crysis Trilogy Remastered brings 4K support, ray tracing, DLSS, improved lighting and textures, and higher quality models of characters and vehicles in the game. It has all been covered in detail in the Digital Foundry video who focused on Crysis 2 comparison across PC and pretty much all gaming consoles.
Crysis-Remastered-Review-Guide.jpg
 
It is not fair at all. Crytek and Saber Interactive are using a hybrid RT approach, so on Turing and Ampere RT gets accelerated unlike RNDA2. Crytek claims it improves performance around 15 to 20%.

But i think it shows how bad RDNA2 as an architecture is. When it comes to compute performance it is barely faster than Pascal on a transistor level. A 3060 can deliver ~60 FPS with maximum RT, a "card designed for 1080p" like the 6600XT is falling from the cliff. So many transistors are spent on something like a L3 cache to hide the fact that RDNA2 is not effcient enough. The best transistor is the one which is doing calculations instead of holding informations.

The other problem for AMD cards in Crysis games is that for some reason they have very bad performance in places and this has been there since Vega. Many people use DXVK to use Vulkan for improved fps.

The remasters are still on DX11 and RDNA2 cards still lag wrt CPU driver usage there.
 
The other problem for AMD cards in Crysis games is that for some reason they have very bad performance in places and this has been there since Vega. Many people use DXVK to use Vulkan for improved fps.

The remasters are still on DX11 and RDNA2 cards still lag wrt CPU driver usage there.

I'm curious.
Did they fixed those infamous "tessellation tricks" on Crysis 2?
 
The performance cost is a valid concern. I feel the same way about native 4K which at tv viewing distances adds almost nothing over 1440p or 1600p. It’s a waste of rendering performance.

The inconsistency is in the standards for what should be considered material visual impact. It’s one thing to say oh that’s nice but I turn it off cause it’s too expensive which is a reasonable stance. It’s another thing to argue that improvements in lighting, shadows and reflections are no big deal. That’s the really confusing thing to hear on a graphics forum.
Some here sure get very upset when others don't agree with their subjective views on how beneficial current RT implementations are. RT just hasn't changed the status quo for me in visuals at this point. The best looking games IMO are all still using rasterization. Most cases of RT take a game that looks anywhere from solid/decent to bad and make it look a little bit better for a massive performance cost. It shouldn't be hard to understand why people aren't thrilled with current RT implementations.

https://www.dsogaming.com/pc-performance-analyses/industria-pc-performance-analysis/
 
Some here sure get very upset when others don't agree with their subjective views on how beneficial current RT implementations are. RT just hasn't changed the status quo for me in visuals at this point. The best looking games IMO are all still using rasterization. Most cases of RT take a game that looks anywhere from solid/decent to bad and make it look a little bit better for a massive performance cost. It shouldn't be hard to understand why people aren't thrilled with current RT implementations.

https://www.dsogaming.com/pc-performance-analyses/industria-pc-performance-analysis/

You're not forced to be thrilled. But there are objective IQ improvements that people go out of their way to poo poo for some reason. And yes Industria is not a great example of RT or anything else for that matter.
 
You're not forced to be thrilled. But there are objective IQ improvements that people go out of their way to poo poo for some reason. And yes Industria is not a great example of RT or anything else for that matter.
Could you point some of these? I seriously haven't noticed anyone downplaying specific RT effects*, just that the overall change in visuals isn't worth the performance loss.

*(Unless you count the pro-current-RT-implementations-folks downplaying effects which don't meet their requirements)
 
Could you point some of these? I seriously haven't noticed anyone downplaying specific RT effects*, just that the overall change in visuals isn't worth the performance loss.

*(Unless you count the pro-current-RT-implementations-folks downplaying effects which don't meet their requirements)

Just a few pages ago we were debating whether or not completely missing transparent reflections in Control were a significant IQ issue.
 
Back
Top