• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

3DMark Introduces Variable Rate Shading Benchmark

Raevenlord

News Editor
Joined
Aug 12, 2016
Messages
3,755 (1.18/day)
Location
Portugal
System Name The Ryzening
Processor AMD Ryzen 9 5900X
Motherboard MSI X570 MAG TOMAHAWK
Cooling Lian Li Galahad 360mm AIO
Memory 32 GB G.Skill Trident Z F4-3733 (4x 8 GB)
Video Card(s) Gigabyte RTX 3070 Ti
Storage Boot: Transcend MTE220S 2TB, Kintson A2000 1TB, Seagate Firewolf Pro 14 TB
Display(s) Acer Nitro VG270UP (1440p 144 Hz IPS)
Case Lian Li O11DX Dynamic White
Audio Device(s) iFi Audio Zen DAC
Power Supply Seasonic Focus+ 750 W
Mouse Cooler Master Masterkeys Lite L
Keyboard Cooler Master Masterkeys Lite L
Software Windows 10 x64
3DMark today announced they've introduced a new benchmarking feature. Specifically developed to test Variable Rate Shading (VRS) performance and image quality differences, the new feature allows users to actually visualize the performance and image quality differences associated with more aggressive (or less aggressive) VRS settings. The algorithm is a smart one - it aims to reduce the number of pixel shader operations on surfaces where detail isn't as important (such as frame edges, fast-moving objects, darkened areas, etc) so as to improve performance and shave some precious milliseconds in the deployment of each frame.

To run this test, you will need Windows 10 version 1903 or later and a DirectX 12 GPU that supports Tier 1 VRS and the "AdditionalShadingRatesSupported" capability, such as an NVIDIA Turing-based GPU or an Intel Ice Lake CPU. The VRS feature test is available now as a free update for 3DMark Advanced Edition, or from now until September 2, 3DMark is 75% off when you buy it from Steam or the UL benchmarks website.






View at TechPowerUp Main Site
 
Nice that Ice Lake supports this, and it's obviously no surprise the much-maligned Turing once again is packing a forward looking feature.

Is miss the days Nvidia was always playing catch up.
 
Turns out VRS doesn't really need dedicated hardware for it and can be implemented purely in software, the new CoD on consoles will have this, of all things. So meh, this isn't all that amazing after all.
 
And now we will be able to compare performance figures, I can definitely see the benefits for the likes of VR, but then Nv have excelled there too.
 

You know, future proof hardware, just you wait until it's untapped potential is unleashed. GCN were compute monsters, Nvidia hardware was "gimped". Blah blah blah.

I'm new to all this, apparently it's all easily done in software anyway.
 
So another image quality cheating to get top ranks in gaming benchmarks, good job, now we will see more "AMD is bad because of ..."
 
IMO VRS ON and OFF visually looks like it just dropped anisotropic filtering and use bilinear
 
The tree right in the middle of the screen on the VRS ON screen looks badly blurred. I guess it's not the best example of this technique.

Edit: actually it's not just one tree - there are many trees with this issue.

Edit 2: also there's something wrong with the leaves on the ground. They look pixellated.
 
VRS is great for peripheral vision, throwing it in the middle of the screen can only work with shallow DOF to mask it.
 
Nice to see the side by sides, gives a good impression. Nice little test.

Whether or not this is good... of course it is. Rendering will always be full of trickery and it already is. As long as your perception of the image is still good, no harm done. That being said, I do see where VRS cuts the corners, unfortunately. Needs careful application.
 
Not gonna lie, the VRS ON image looks like shit.
 
Just to have a glimpse at actual implementation of VRS:
 
So another image quality cheating to get top ranks in gaming benchmarks, good job, now we will see more "AMD is bad because of ..."

and what did you think AMD Rapid Packed Math did? and back then we saw people said we need more tech like this where it can improve performance.
 
Would it kill them to make the test look good?
 
Well it reduces strain on the system...by reducing the quality of the image, very noticeably so, so I dont really see any value in this tbh.
 
On a related note...
Nvidia Adaptive Shading in Wolfenstein:Youngblood


Yeah seems interesting, free performance boost that unless you really like to pause and zoom in whilst gaming, you won't notice a thing frankly.

 
Well it reduces strain on the system...by reducing the quality of the image, very noticeably so, so I dont really see any value in this tbh.
Reminds me of FXAA and DLSS.

Yeah seems interesting, free performance boost that unless you really like to pause and zoom in whilst gaming, you won't notice a thing frankly.

It's impossible to tell anything from a lossy video unless it's so bad that the compression doesn't mask it.
 
Funny, that is exactly what people are doing in this thread.

Mind blown.
 
Actually, I think with VRS off the image looks better (sharper). Especially in the farther scene.
 
Actually, I think with VRS off the image looks better (sharper). Especially in the farther scene.

yeah, I think you should read the post again :P
VRS is there to increase performance at HOPEFULLY no image quality cost.

But as we can, I think, all see, the image quality loss is very noticeable.
 
yeah, I think you should read the post again :p
VRS is there to increase performance at HOPEFULLY no image quality cost.

But as we can, I think, all see, the image quality loss is very noticeable.
That was the point of my post. even though it shouldn't decrease the quality in fact it does and it is very noticeable.
 
Back
Top