Watch Dogs Legion Benchmark Test & Performance Analysis

[Luminescent]

On PC - you need a 1000$ GPU to play the game maxed out
Console - PS5 400$ and you are good to go
This is insane, both Amd and Nvidia are charging way too much money for a GPU.

 

[ZoneDymo]

Dear lord these comments are pathetic….like...why are you guys even here....

Take a look at (EDIT) Digital Foundry's recent vid on watchdogs, you might learn something and appreciate stuff a bit more...

And lets not forget that "graphics" are completely arbitrary, this is purely a mindgame where people think "if my latest 800 dollar gpu cant do Ultra on this game then clearly the game is dogsh*t", seemingly not realizing that the devs could have just as easily called their current medium settings Ultra and voila, eveyone can max it! praise the devs right?
Thought we all appreciated Crysis nowadays, some seemingly took a bit longer to understand this but in the end we are all ok with it pushing the boundries right?

Here is a game that pushes something we all want to see, progress, ray tracing, and then they get scolded because it "shockingly" is pretty heavy to run....its almost as if Real Time Ray Tracing hasnt been a thing until now because its heavy to run or something….

And hell this is only reflections AND those are even aided/suplimented by standard Space Screen Reflections, we are not even talking shadows and actual lighting yet here.

On PC - you need a 1000$ GPU to play the game maxed out
Console - PS5 400$ and you are good to go
This is insane, both Amd and Nvidia are charging way too much money for a GPU.

This comment is beyond silly, how did you end up on this forum with such a rediculous mindset.
Does anyone really have to explain this to you? seriously?

 

[s3thra]

Dear lord these comments are pathetic….like...why are you guys even here....

Take a look at Global Foundry's recent vid on watchdogs, you might learn something and appreciate stuff a bit more...

And lets not forget that "graphics" are completely arbitrary, this is purely a mindgame where people think "if my latest 800 dollar gpu cant do Ultra on this game then clearly the game is dogsh*t", seemingly not realizing that the devs could have just as easily called their current medium settings Ultra and voila, eveyone can max it! praise the devs right?
Thought we all appreciated Crysis nowadays, some seemingly took a bit longer to understand this but in the end we are all ok with it pushing the boundries right?

Here is a game that pushes something we all want to see, progress, ray tracing, and then they get scolded because it "shockingly" is pretty heavy to run....its almost as if Real Time Ray Tracing hasnt been a thing until now because its heavy to run or something….

And hell this is only reflections AND those are even aided/suplimented by standard Space Screen Reflections, we are not even talking shadows and actual lighting yet here.



This comment is beyond silly, how did you end up on this forum with such a rediculous mindset.
Does anyone really have to explain this to you? seriously?

Thank you!

I’ve always thought of “Ultra” settings for a game to be the absolute limits of that game which push the boundaries of the game engine. Great for reviews like this, and meant to make the latest cards absolutely bleed and push the industry along.

In real world practice, “High” or “Medium” is what you should be looking at playing at, depending on your card. In a lot of the games I play the visual difference between “High” and “Ultra” is negligible, but the performance difference can be massive depending on the game, and that’s how it should be for the reason I stated in my first paragraph.

In PC gaming we have the option to tweak settings based on our configuration, it’s not just a matter of blasting every setting to the max and expecting everything to run perfectly fine.

 

[Prince Valiant]

I see DLSS is still smearing jelly on graphics. Dismal performance for how mediocre the models look.

 

[B-Real]

Interesting results. The game has built-in benchmark. I think you used that Wizzard. How did Hilbert at Guru3D got 65 fps average with the 3090 while you got only 56? That's a 16% difference which is huge. Also, RX 5700 was beating the RTX 2070 in every resolution both on very high and ultra by 3 to 9% (on 4K, they were on par) while in your benchmarks, the RTX 2070 was 5, 11 and 15% faster than the RX 5700 in the 3 resolutions

 

[W1zzard]

Interesting results. The game has built-in benchmark. I think you used that Wizzard. How did Hilbert at Guru3D got 65 fps average with the 3090 while you got only 56? That's a 16% difference which is huge.

Cmon .. please check the conclusion

Edit: Amazed how people can still miss this. I made extra sure this time to avoid such comments .. it's also mentioned in each of the charts

 

[Dredi]

I really don't like TAA. It's even more controversial than DLSS.

DLSS is just a (fairly ok) TAA implementation.

 

[B-Real]

Cmon .. please check the conclusion

Edit: Amazed how people can still miss this. I made extra sure this time to avoid such comments .. it's also mentioned in each of the charts

Ok, you had the high res pack, they didn't. Got it, ty.

 

[W1zzard]

Ok, you had the high res pack, they didn't. Got it, ty.

Nope, wrong answer. Try again

 

[dirtyferret]

Ok, you had the high res pack, they didn't. Got it, ty.

While other publications are testing using the integrated benchmark, I found that to be way too optimistic for what actual gameplay runs at. For example, with GeForce RTX 3080 at 1080p, the integrated benchmarks runs at 106 FPS, while actual gameplay sits at around 85 FPS—a huge difference. Now, you of course could argue that not all gameplay happens in the open world, and that indoors areas run higher FPS

 

[Steevo]

54 fps on $1500 GPU? What a joke! No matter how fast hardware gets, it can't save us from abysmal port running like S***.

The "port" is running on X86-64 hardware. Perhaps the new 5000 series Ryzen CPUs and 6900XT will perform better with their new cache..

 

[Luminescent]

Take a look at Global Foundry's recent vid on watchdogs, you might learn something and appreciate stuff a bit more...


This comment is beyond silly, how did you end up on this forum with such a rediculous mindset.
Does anyone really have to explain this to you? seriously?

You mean Digital foundry ?
Explain why my comment is silly ? beyond the bullshit we are all enthusiasts here and willing to throw money at Nvidia and Amd and be dazzeled by reflections and oh my how good are the graphics.
They are not, they are still cartoonish and all the marketing crap mean nothing when you can get the same thing for 400$ in one place and for 3000$ in another place (decent desktop PC)
It's just that simple, consoles are superior to PC's for games, cheaper, better, much more games and better.
Buy a console for games and buy a PC for work, a "gaming" GPU costs more than a console.

 

[Steevo]

Cmon .. please check the conclusion

Edit: Amazed how people can still miss this. I made extra sure this time to avoid such comments .. it's also mentioned in each of the charts

Words are hard. Harder for some.

 

[SKD007]

Are you playing the integrated benchmark?

That is correct my friend. I did. My Ram is 3200Mhz 14c. not sure if that is helping.

 

[W1zzard]

That is correct my friend. I did. My Ram is 3200Mhz 14c. not sure if that is helping.

What's your FPS when standing on a street?

 

[Akkedie]

I think should did a ultra settings dx12 vs dx11 comparison of fps. when i was doing settings on mine, dx12 gave me 68fps and dx11 ran at 72fps. Sure people would say things in dx12 were active over 11 giving the fps boost but included in benchmark is a fps graph and it was very telling to say least. On the left of the image is dx11 you can see its tight and compact, dx12 on other hand has a lot of latency in frames.

CPU performance data in the benchmark is 100% useless. Gotta be in the world to see anything useful.

 

[rtwjunkie]

I see DLSS is still smearing jelly on graphics. Dismal performance for how mediocre the models look.

Careful. The last time I talked about DLSS looking blurry compared to some other AA options the pro-DLSS Brigade attacked me in force.

 

[InVasMani]

Without watching the video, either the implementation is broken in your example or you are not actually enabling AF. Do a quick Google search and do some reading or go download the old 3DCenter AF Tester and play with it a bit. AF is a method of improving image quality by pushing out the boundary of the closest mipmap and smoothing out the blending between the successive levels of mipmaps as the textures recede from the image plane.

Even if you don't believe me or literally every article or bit of documentation that exists on the subject for some odd reason, ask yourself why anyone would enable an option the decreases visual quality (what you are implying) while simultaneously negatively impacting performance. It doesn't make any sense.

MFAA requires half as much performance as MSAA at the same settings and reduces texture crawl and shimmering. Downsampling yes that requires more performance however it also is better than post process AA techniques in general it doesn't distort things like shading, lighting, and reflections the way many post process AA methods can. Negative LOD bias makes brings it a ton fine detail to a texture from using sharper mipmap quality. That introduced shimmer/texture crawl of course, but MFAA reduces so they kind of cancel one another out a fair degree all while and still at less performance than MSAA on it's own with better quality subjectively speaking. The issue is mostly the texture shimmering/crawl in motion AF most certainly helps with that, but at the expense of texture detail taking a digger the scene blurs more and deep color quality seems to noise dive a little bit.

MFAA uses half the performance of MSAA at the same setting value. It also simultaneously reduces texture shimmer and crawling as a nice side effect. As for some downsampling because it's not a post process technique it doesn't have any of the associated subjective negatives of post process AA's techniques. More downsampling with less post process AA isn't a terrible option and consideration taking that into account. You can lose details on shading, lighting, and reflections with various AA techniques so downsampling is actually quite optimal though some AA is helpful, but keeping it more bare minimum is a good practice weigh the pro's and con's of the technique applied in conjunction with other settings involved. In the case of MFAA and negative LOD bias they pair well together. In the case of AF with both of those things if you use a higher MFAA value with a higher negative LOD bias you can get away with using a lower AF value with less perceived degradation and w/o introducing over sharpening superellipse issues that are present with x8 and x16AF settings with negative LOD bias set more aggressively like -3.0 value. It's most noticeable with x16 AF and a more extreme negative LOD bias value -3.0 to -1.5 value. It's not as pronounced at x8AF and at x4AF it's hard to tell.

It's worth noting that if you sharpen a texture you often bring out a ton of fine detail within the texture especially with bump mapping and course shadows, lighting, and reflections are more apparent with better sharpness. Color seems to be more accurate and rich with lower AF settings slightly it gets more blurry and washed out as you increase the AF values. On the one hand you fix the shimmering/crawl in motion on the other hand you killing more stationary color and texture detail with AF being set increasingly higher value. There is a reason I like MFAA performance is good and it also decrease and offsets texture crawl and shimmering which allows for more flexibility with the AF value you set. I think AF is generally great up to x4 and sometimes you have to compare from game to game between x2 and x4 on how it impacts those distant oblique angles relative to the other trade offs. If your using a higher MFAA value with negative LOD bias of -3.0 to -1.5 you might want to consider playing around with the AF value a bit and comparing. I mean fixing shimmer in motion to make the far texture quality more blurry is a trade off. MFAA is good compromise if you utilize negative LOD bias values which helps negate the need for setting AF higher while increasing fine texture detail in a readily obvious manner. A higher value of AF has positives and negatives with better texture qualities, bump mapping, r higher resolution and DPI, color reproduction, and negative LOD bias it becomes more noticeable and a bigger relative concern.

 

[mahoney]

Dear lord these comments are pathetic….like...why are you guys even here....

Take a look at Global Foundry's recent vid on watchdogs, you might learn something and appreciate stuff a bit more...

And lets not forget that "graphics" are completely arbitrary, this is purely a mindgame where people think "if my latest 800 dollar gpu cant do Ultra on this game then clearly the game is dogsh*t", seemingly not realizing that the devs could have just as easily called their current medium settings Ultra and voila, eveyone can max it! praise the devs right?
Thought we all appreciated Crysis nowadays, some seemingly took a bit longer to understand this but in the end we are all ok with it pushing the boundries right?

Here is a game that pushes something we all want to see, progress, ray tracing, and then they get scolded because it "shockingly" is pretty heavy to run....its almost as if Real Time Ray Tracing hasnt been a thing until now because its heavy to run or something….

And hell this is only reflections AND those are even aided/suplimented by standard Space Screen Reflections, we are not even talking shadows and actual lighting yet here.



This comment is beyond silly, how did you end up on this forum with such a rediculous mindset.
Does anyone really have to explain this to you? seriously?

Ubisoft aka Ubishit has a track record of poor optimizations not to mention the shit they created with the drm back in the late 00s - securom...

More trouble for Ubisoft's DRM system

Users of Ubisoft's PC games Assassin's Creed II and Silent Hunter 5 have been unable to play over the weekend as the se…

www.gamesindustry.biz

They are the absolute worst at it.

 

[SKD007]

What's your FPS when standing on a street?

on what res and settings ? 1440p Ultra ? Or 4K ultra

 

[ZoneDymo]

You mean Digital foundry ?
Explain why my comment is silly ? beyond the bullshit we are all enthusiasts here and willing to throw money at Nvidia and Amd and be dazzeled by reflections and oh my how good are the graphics.
They are not, they are still cartoonish and all the marketing crap mean nothing when you can get the same thing for 400$ in one place and for 3000$ in another place (decent desktop PC)
It's just that simple, consoles are superior to PC's for games, cheaper, better, much more games and better.
Buy a console for games and buy a PC for work, a "gaming" GPU costs more than a console.

Yes I meant digital foundry, my bad, global foundry is the fab, but ok....

You said:
On PC - you need a 1000$ GPU to play the game maxed out
Console - PS5 400$ and you are good to go
This is insane, both Amd and Nvidia are charging way too much money for a GPU.

If you think that the consoles will produce image quality remotely close to running the game MAXED OUT, then idk what to tell you….they just arnt.
If the game was benchmarked at a mix of settings with stuff mostly on med, the fps numbers would be way higher.
Maxed Out is in another world then the console settings but obviously if you dont care that much for Visuals and running a game above 30 fps then by all means get a console, far smarter choice.

This is not a dig at consoles at all, its just a completely weird comparison to make, my old af pc is also "good to go" to play this game, just have to turn some settings down, just like the consoles do.

 

[W1zzard]

on what res and settings ? 1440p Ultra ? Or 4K ultra

yeah ultra, ideally both, if you can

 

[InVasMani]

What was the deal with the reflections on the floors with the DLSS on/off comparison. It appeared like the DLSS made some of them vanish byproduct of that post process AA technique itself or did RTRT settings get tampered with as well mistakenly!?

 

[W1zzard]

What was the deal with the reflections on the floors with the DLSS on/off comparison. It appeared like the DLSS made some of them vanish byproduct of that post process AA technique itself or did RTRT settings get tampered with as well mistakenly!?

Great question, now I see it too. I didn't change any RT settings, it was off for all the DLSS screenshots. Maybe the light source is time dependent and has changed in the between changing settings and screenshot.

 

[InVasMani]

Disabling TAA with DLSS seems like it would make sense to consider if testing DLSS's image quality.