The Denuvo DRM Thread

[Morpheus Kitami]

Denuvo definitely impacts performances. It will make a game that would otherwise run fine in the most powerful rig to sometimes sutters and had fps drop.

I've found drastic examples like that - AC Origins or one of the RE games - but in all of them the devs combined Denuvo with their own DRM, resulting in a literal shit show.

I mean the first Denuvo implementation was in the first Lords of the Fallen which I've bought and enjoyed without any performance problems. And that was 10 years ago, with a 13 yo CPU.

I think usually the impact of Denuvo on a modern CPU is pretty negligible IF (and only if) the devs implement the checks in a way that they aren't called super frequently (IIRC Ubisoft was calling one of the DRM checks on every frame with AC Origins).

On older CPUs it definitely seems to have a more noticeable impact. In addition there are also cases where the devs implement the checks in the worst possible ways, Capcom and Ubisoft seem to be guilty of this with alarming frequency.

All of this said, regardless of whether the impact is negligible, it's still an impact and it's still bullshit that people that pay for a game should have an inferior experience because of an anti-piracy measure.

Denuvo could also do far more to prevent their customers from implementing the DRM in a subpar manner if they actually cared about their reputation with gamers. They could easily have a few devs who liaise with studios on a more frequent basis to ensure that the DRM is being implemented in an optimised manner, but this would require both a level of effort and integrity on their part, and they have neither.

That's an interesting observation, but consider how many developers these days are guilty of having shit optimization? It's a dead horse at this point to say that they just dump a tangled mess on your system. Because hey, we have more computing power than ever and if people can't play it now, they can when they upgrade their computer/console. It's not just Capcom and Ubisoft, IIRC, the Injustice games were awful with it and I'm sure I'm just out of examples because I don't play many new games. Has anyone actually done it well? Maybe that cricket game that was infamous for never being cracked, which would be a pretty hilarious thing to see. The one game nobody interested in the subject played and it's the only one that actually worked.

 

[cvv]

Has anyone actually done it well?

Define well.

Because that's the problem - it's hard to measure how a Denuvo protected game would've run without it.

In some cases people benchmarked a Denuvo protected game and then Denuvo was removed by the devs so the game was benchmarked again. Which was the case with the 2016 Doom for example. And which is completely useless because there was half a year between the tests and because the game kept getting updated and optimized plus Nvidia was putting out optimized GPU drivers too.

Sometimes the devs announce Denuvo removal in advance so people can benchmark the game in the span of a few days. Which is what they did with Hitman, Abzu, Mad Max (all no difference in performance), Sniper (difference of about 4 FPS) and ME Andromeda (difference of about 15 FPS).

That's why this is such a rabbit hole. It's virtually impossible to gauge the real general impact Denuvo has on performance. From everything I've seen it seems it completely depends on the implementation. If it's done competently, Denuvo doesn't impact performance at all. But if you're Bioware...

 

[Tacgnol]

The other issue is that it can have an impact on things beyond straight fps comparisons. With Metro Exodus for instance, there was an impact on loading times between the protected and unprotected versions.

Like everything, it seems to come down to how well the devs have implemented it.

 

[Raghar]

people claiming cracked Denuvo games perform way better than uncracked, not realizing Empress wasn't actually removing or disabling Denuvo

People also compared games after denuvo was officially removed from them

Is there any credible, objective analysis of this? Something Digital Foundry or some such did? I'd actually love to read up on this. Because most of what I have found was of the "trust me bro" kind.

Yea there is. But you can't read it because of your ignore list.

Anyway there IS massive performance AND security hit because crossing security layer between OS rings does create MASSIVE penalty. It can increase latency, but it's at least 5 percent of CPU time nowadays.

Even larger problem is that DRM allows shutting down stuff that a person using in his private home on his own property. And that's really bad because it kills one of basic premises of PC that allowed unpaid research in free user's time at theirs homes. Which was kinda important for high skills of the society and was one of driving force for tech advancement in 90s.

 

[Arthandas]

If it's done competently, Denuvo doesn't impact performance at all.

It literally always affects performance because it's impossible for a 400MB exe to outperform a 20MB exe. Even if it doesn't directly lower the framerate, it will affect loading times, framepacing, asset streaming and other shit.

RE Village literally stuttered every time you killed an enemy and some effects completely tanked the framerate because of Denuvo. Tested and confirmed by Digital Foundry.

Harada confirms Tekken 7 stutters because of Denuvo (also, that's the reason there's no Denuvo in T8).

DMC 5 runs around 7% slower thanks to Denuvo. Tested and confirmed by Digital Foundry.

And those are just 3 examples from the top of my head. There were also day 1 denuvoless exe leaks for some Persona games and Metaphor ReFantazio, I'm sure if you tested those builds you would get similar results as above.

 

[cvv]

And those are just 3 examples from the top of my head.

Yes. And there are many examples of games with and without Denuvo running exactly the same.

Which points to a problem with implementation.

Which was already discussed.

So what exactly are you objecting to?

 

[Arthandas]

And there are many examples of games with and without Denuvo running exactly the same.

Like? I doubt you can provide a single proper comparison showing a Denuvo game having the same framerate, framepacing, loading times and no additional issues like a denuvoless version.

 

[soutaiseiriron]

Anyway there IS massive performance AND security hit because crossing security layer between OS rings does create MASSIVE penalty. It can increase latency, but it's at least 5 percent of CPU time nowadays.

denuvo operates entirely in usermode, though. it's not a kernel driver, it's not safedisc or starforce.
this discussion was had like 10 pages ago. it has an impact on 1% lows and loading times. how bad the impact on lows is, is dependent on how well profiled the game is. a well profiled denuvo game has relatively negligible performance impact.

 

[Gerrard]

Imagine actually being a nigger defending this malware.

 

[cvv]

And there are many examples of games with and without Denuvo running exactly the same.

I doubt

An argument from personal incredulity, very cool.

 

[Prince Ketchup]

It would have been fine if Denuvo had been used in games for a year or two and then dropped, but some companies don't ever remove it. And you have EA, which has DRM still active in games like BF4/Titanfall 2.

 

[Arthandas]

a) The guy has literally no idea how to measure framepacing, the way he does it is retarded and irrelevant. Both games need to be running at the same exact stable fps (for example 60) and then you measure how close they match the proper refresh rate (16.6ms for 60 fps).
b) They made a follow-up with loading times. To summarize both videos:

Hitman - only fps tested
Abzu - more than double loading times with denuvo
Sherlock Holmes - lower framerate with denuvo
Mad Max - almost double loading times with denuvo
Agents of Mayhem - lower framerate with denuvo in 2 out of 3 scenes + slightly longer loading times
Sniper - lower framerate with denuvo
Rime - 50% longer loading times with denuvo

So, thanks for proving my point?

 

[cvv]

a) The guy has literally no idea how to measure framepacing, the way he does it is retarded and irrelevant.

- A: give me proof!!1!
- B: gives proof
- A: not that proof!1!!

i'm shock so unfair

 

[Arthandas]

cvv proving to us denuvo has no impact on performance by linking a video where half of the games have straight out lower framerates and 50-100% longer load times tested in a follow-up clip he conveniently ignored:

 

[Hirato]

people claiming cracked Denuvo games perform way better than uncracked, not realizing Empress wasn't actually removing or disabling Denuvo

People also compared games after denuvo was officially removed from them

Is there any credible, objective analysis of this? Something Digital Foundry or some such did? I'd actually love to read up on this. Because most of what I have found was of the "trust me bro" kind.

As I've explained many times, including this thread, games do not scale linearly.
Making a game use as little as 2% more CPU power for the same result can literally half your framerate.

This is why every benchmark result for this runs the gamut from "no discernible impact" to "crippling stutter" or "overall 30% drop in FPS".
The people who benchmark this stuff always produce GPU benchmarks, and not CPU benchmarks, so any actual numbers we have around "FPS" are all bullshit.
The one they don't mess up are the loading times, and those are always greatly crippled relative to the no-denuvo versions.

 

[Odoryuk]

Daily reminder that unbroken Denuvo games will turn into a pumpkin once the servers will eventually go down

 

[hajro]

Daily remainder that germanic people arent people

 

[Azdul]

people claiming cracked Denuvo games perform way better than uncracked, not realizing Empress wasn't actually removing or disabling Denuvo

People also compared games after denuvo was officially removed from them

Is there any credible, objective analysis of this? Something Digital Foundry or some such did? I'd actually love to read up on this. Because most of what I have found was of the "trust me bro" kind.

Yea there is. But you can't read it because of your ignore list.

Anyway there IS massive performance AND security hit because crossing security layer between OS rings does create MASSIVE penalty. It can increase latency, but it's at least 5 percent of CPU time nowadays.

Even larger problem is that DRM allows shutting down stuff that a person using in his private home on his own property. And that's really bad because it kills one of basic premises of PC that allowed unpaid research in free user's time at theirs homes. Which was kinda important for high skills of the society and was one of driving force for tech advancement in 90s.

There SHOULD be massive performance hit when crossing the layers between OS rings. All the caches should be purged, so passwords could not be retrieved by JavaScript running on a web page and measuring response times: Meltdown Arxiv Spectre Arxiv

However,


Good thing you should be safe if you have nothing to hide.

 

[Azdul]

Daily remainder that germanic people arent people

Blaming all germanic people is racist.

First of all, Denuvo was created by Austrians. While some people say that they are the missing link between Bavarians and humans, most well-known Austrians are respected all around the world:

 

[Tacgnol]

Always used to be a good way of trolling Australians.