Thursday, December 5th 2019

UL Benchmarks Outs 3DMark Feature Test for Variable-Rate Shading Tier-2

UL Benchmarks today announced an update to 3DMark, with the expansion of the Variable-Rate Shading (VRS) feature-test with support for VRS Tier-2. A component of DirectX 12, VRS Tier 1 is supported by NVIDIA "Turing" and Intel Gen11 graphics architectures (Ice Lake's iGPU). VRS Tier-2 is currently supported only by NVIDIA "Turing" GPUs. VRS Tier-2 adds a few performance enhancements such as lower levels of shading for areas of the scene with low contrast to their surroundings (think areas under shadow), yielding performance gains. The 3DMark VRS test runs in two passes, pass-1 runs with VRS-off to provide a point of reference; and pass-2 with VRS-on, to test performance gained. The 3DMark update with VRS Tier-2 test will apply for the Advanced and Professional editions.

DOWNLOAD: 3DMark v2.11.6846
Add your own comment

58 Comments on UL Benchmarks Outs 3DMark Feature Test for Variable-Rate Shading Tier-2

#1
deu
Soo... The first two pictures are 110% the same only with on/off in the bottom? :D
Posted on Reply
#2
The Quim Reaper
67% uplift in framerates when I tested it (2080) @4K !

If the aim is to keep framerates above 60 @ 4K at all times, then it works very well at doing that.

..I'm sure if you looked hard enough you'd see where the cutbacks are but it's not obvious when in motion.
Posted on Reply
#3
ShurikN
The Quim Reaper
..I'm sure if you looked hard enough you'd see where the cutbacks are but it's not obvious when in motion.
Man, I've been looking at it for 3 minutes, fast switching with arrow keys, and I cannot see any differences.
Whether it's different in motion or not I can't say, but the stills look identical.


Is this a feature that needs to be developed for by the game devs, or is this something that you just go into NV control panel and flip a switch?
Posted on Reply
#4
FM_Jarnis
ShurikN
Man, I've been looking at it for 3 minutes, fast switching with arrow keys, and I cannot see any differences.
Whether it's different in motion or not I can't say, but the stills look identical.


Is this a feature that needs to be developed for by the game devs, or is this something that you just go into NV control panel and flip a switch?
Needs to be developed into the game engine.
Posted on Reply
#5
HTC
ShurikN
Man, I've been looking at it for 3 minutes, fast switching with arrow keys, and I cannot see any differences.
Whether it's different in motion or not I can't say, but the stills look identical.


Is this a feature that needs to be developed for by the game devs, or is this something that you just go into NV control panel and flip a switch?
If you look closely, there's an "off" in one picture and an "on" in another: bottom right corner ...
Posted on Reply
#6
Hyderz
i had a 3 mins look to compare and i've noticed with on pic there is less texture tearing..
1. on the power box there is some text there its still blurry but its just a little bit sharper? than the off pic
2. if you look at the buildings in the background from the top left in the gap where u see the building its noticeably sharper
3. near the car where there is a patch of smooth darker dirt its ever so slightly sharper
4. the wooden power pole near the car is noticeably sharper
Posted on Reply
#7
Xaled
The Quim Reaper
67% uplift in framerates when I tested it (2080) @4K !

If the aim is to keep framerates above 60 @ 4K at all times, then it works very well at doing that.

..I'm sure if you looked hard enough you'd see where the cutbacks are but it's not obvious when in motion.
Did you test it in a game where 67% of it is covered by shadows and are dark areas?
Posted on Reply
#8
RH92
The Quim Reaper
67% uplift in framerates when I tested it (2080) @4K !
That's great , now obviously results will varie depending on how well the tech is implemented into individual games which ultimately depends on game developpers but yeah any performance uplift would be great for those who game in high resolutions !
Posted on Reply
#9
delshay
I notice a clearer dirt mark below the branch on the wall. On it is clear, & off you can just make it out. Look just above the dark patch on the wall, first there is one dotted dirt mark, then there is two "when on"..
Posted on Reply
#10
Prince Valiant
ShurikN
Man, I've been looking at it for 3 minutes, fast switching with arrow keys, and I cannot see any differences.
Whether it's different in motion or not I can't say, but the stills look identical.


Is this a feature that needs to be developed for by the game devs, or is this something that you just go into NV control panel and flip a switch?
The building to the right shows an obvious drop in detail even with the lossy and resized screenshots here.

Here's the UL post (with lossless screenshots):
https://benchmarks.ul.com/news/new-ways-to-test-variable-rate-shading-with-3dmark
Direct links:
https://benchmarks.ul.com/hwc/tmp/3dmark-vrs-feature-test-tier2-screenshot-vrs-off.png
https://benchmarks.ul.com/hwc/tmp/3dmark-vrs-feature-test-tier2-screenshot-vrs-on.png

The quality dip is as obvious as it was on the previous VRS test.
Posted on Reply
#11
the54thvoid
I've had a look at both, full-size images and I cannot see what you guys are on about? I don't question your own vision or your subjective feel but the differences, if any, are exceptionally small. Can you do a full size, side by side comparison of the parts that have been mentioned that are 'very obvious'? It would help to show me what you can see that I cannot. Given that places in shadow, should receive lower texture resolution (because that's how human vision actually works), the differences would have to be profound - and that is not the case.
Posted on Reply
#12
Maelwyse
the54thvoid
I've had a look at both, full-size images and I cannot see what you guys are on about? I don't question your own vision or your subjective feel but the differences, if any, are exceptionally small. Can you do a full size, side by side comparison of the parts that have been mentioned that are 'very obvious'? It would help to show me what you can see that I cannot. Given that places in shadow, should receive lower texture resolution (because that's how human vision actually works), the differences would have to be profound - and that is not the case.
I think that's the point. If I read this right, the desire from development is that this should be low-impact to visual quality, but a substantial improvement in framerates. The wall on the right side certainly has a small texture quality loss, but if it were to give me a noticeably more reliable framerate, or a noticeably improved framerate, I think it would be a win to use this.

And since I'm about to plan to build a new PC, specifically targeting framerates, and have already been willing to drop from 4K@60 down to 1440p 144hz, I'm EXACTLY the targeted demographic.
Posted on Reply
#13
Prince Valiant
the54thvoid
I've had a look at both, full-size images and I cannot see what you guys are on about? I don't question your own vision or your subjective feel but the differences, if any, are exceptionally small. Can you do a full size, side by side comparison of the parts that have been mentioned that are 'very obvious'? It would help to show me what you can see that I cannot. Given that places in shadow, should receive lower texture resolution (because that's how human vision actually works), the differences would have to be profound - and that is not the case.
I've highlighted areas to focus on but it affects almost the entire section I've cropped:







Other areas:
The other junction box, the gas can, the hood of the car, the power pole, the walls, the window shutters, the trashbag, the columns, the ground texture in some spots.
Posted on Reply
#15
the54thvoid
Ah, right.

Okay, I see what you mean. However, I'd suggest that it has zero effect whatsoever on visual experience, as the scene in it's entirety is well resolved with excellent detail. This sort of IQ reduction would only annoy gamers that like to put their quests aside and stare at walls. Proverbial paint drying stuff. Given that I had to have it pointed out to me, it's a clear example of an excellent sacrifice for higher fps.
Posted on Reply
#16
Nater
I absolutely CANNOT see a difference on my 1920x1080 monitor here at work. Even on the original images. Only when I click to zoom in 50% or so and flip back and forth. Texture on the telephone pole is probably the most telling.
Posted on Reply
#17
Prince Valiant
the54thvoid
Ah, right.

Okay, I see what you mean. However, I'd suggest that it has zero effect whatsoever on visual experience, as the scene in it's entirety is well resolved with excellent detail. This sort of IQ reduction would only annoy gamers that like to put their quests aside and stare at walls. Proverbial paint drying stuff. Given that I had to have it pointed out to me, it's a clear example of an excellent sacrifice for higher fps.
I disagree but I'm not bothered if you or anyone else want to use it. As long as it's optional.
Posted on Reply
#18
biffzinker
Looked indistinguishable to me between on and off even with the slow camera panning from left to right.
Posted on Reply
#19
Nater
Prince Valiant
I disagree but I'm not bothered if you or anyone else want to use it. As long as it's optional.
Knowing nVidia this is something they'll turn on by default and be all "LOOK, WE'RE 60% FASTER THAN AMD" and your average hardware review site will completely miss it.
Posted on Reply
#20
InVasMani
Seems like a huge amount of performance uptick that could be used toward post process that could probably end up improving shadows more perceptibly than what was lost. From what I saw of the two images I actually felt the variable rate on looked better than it did with it turned off, but only looked back and fourth for a minute or two. The shadows just looked a bit softer and more real to me rather than overly sharpened and dark at least relative to the drawn scene's time of day it looked more in place. Still 70% more performance roughly is a pretty big deal. It's strange though I wonder where all that added performance derives from primarily is it mostly GPU resources based uplift or is this more a case of less CPU overhead with computation aspects of GPU shading maybe some of both? I'm just thinking this variable shading probably has to benefit both CPU and GPU bound games. It seems like really great tech for the high refresh rate crowd and with real time ray tracing at least when RTRT becomes more relevant and generally acceptable quality and performance still it's a step toward it.
Posted on Reply
#21
Fluffmeister
Turing is an impressive architecture, but it's up to devs to take advantage... on the PC at least.
Posted on Reply
#22
Hotobu
I love this. It's a brilliant idea and I hope they continue to push this further. With hardware giving diminishing returns, software solutions like this are a welcomed addition. Too bad it has to be implemented in development though, and maybe the level of implementation will be adjustable.
Posted on Reply
#23
Vayra86
This is pretty impressive. Bring it on.

Nater
Knowing nVidia this is something they'll turn on by default and be all "LOOK, WE'RE 60% FASTER THAN AMD" and your average hardware review site will completely miss it.
And then a gen later AMD copies it.
Posted on Reply
#24
cucker tarlson
I like the idea of the things nvidia is trying out for the user to have the option to somehow "cheat" the framerate and provide tools for trading a little bit of IQ for noticeable performance gains.

First,there was multi res shading in shadow warrior 2,I usually ran the balanced setting that rendered 20% on each side at 1080p and the central 60% at 1440p.Noticeable in static image,not noticeable during gameplay.the game ran at pretty high framerates anyway.

then dlss,which it seems is really hard to do properly.played the whole control with dlss on and while for the most part it's doing an amazing job of keeping the IQ reasonably high,it's still got frequent glitches and bugs too where it takes too long to load when entering a new area or just plain doesn't recreate some objects.it's really good when it's doing what it's supposed to do,the problem is how inconsistent it is.when comparing static shots 960p dlss produces a tiny bit less jagged image and much less flickering than 1080p while still delivering about 10% more performance.but still,more bugs to work out than most of us would like to see.

with vrs it's easier and the performance gain is almost as high.seems like the best of all three by far.
Posted on Reply
#25
FM_Jarnis
Hotobu
maybe the level of implementation will be adjustable.
I would be surprised if it was not adjustable in games that implement it. In VRS Tier 2 test you can already do this in the Interactive mode that allows you to pause and move along the timeline, plus modify the shading quality, take screenshots... all kinds of things to see how varying the settings affect quality and framerate.
Posted on Reply
Add your own comment