-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
Rumour: Denuvo causes 30-40% additional CPU usage in Assassins Creed: Origins
- Thread starter Poetic.Injustice
- Start date
Älg; said:
Come on, it's not a fucking rumor. Anyone who understands how VMProtect/Denuvo work can make a educated guess about the performance implications. The code that runs under VMProtect is both encrypted and runs in a virtual machine (which has a entirely different instruction set compared with x86-64).
So any piece of obfuscated code that needs to run by the CPU :
- first it needs to be decrypted.
- after that it needs to be interpreted (by converting it between 2 separate instruction sets , the obfuscated VM instruction set and x86-64).
- for AC:O specifically, imagine this being done TWICE : once for vmprotect and immediately afterwards for Denuvo.
This IS SLOW compared with native unobfuscated code. Imagine this needing to be done in the game loop when every frame needs to be drawn (just like it was happening in Rime for instance).
Regarding proof, NOBODY can actually prove anything because for a irrefutable proof you will need access to the obfuscated binary (as it spit out by the MSVC compiler) to benchmark between the 2 modes. And , obviously, only Ubisoft has access to both binaries.
No you can't, because you have no idea how much of an impact this would have on performance in total since you have no idea how much code is actually obfuscated and how time critical it is.
If you say that performance drops by 30%, then that means you know exactly what is going on. Or you're talking out of your ass.
And of course they could benchmark the game as soon as they have cracked it, but they won't. We only have benchmarks of games for which it was deliberately removed by the developers like Doom and they showed nowhere near as much of a difference (and what there is might even be due to optimizations).
Virtually no-one here understands how they work though - and even people that do understand the technical details are throwing out assumptions based on zero context (and not knowing how Denuvo actually uses virtualization).Aklamarth said:
The fact that we have "news" talking about stacking the two is just exemplifying the confusion of internet-communities at the concept. It's analogous to the pre-internet era of people zipping their rar archives but at least excuse there was pre-internet (for most part).
Yea but DRM companies are guilty by association, so lack of proof in the other direction, is just as good as.dr_rus said:
deadscreensky
Member
Yeah, we've heard so many fictional horror stories over the years about DRM sapping performance that it's hard to take this latest tale -- naturally delivered with no evidence or proof! -- seriously.
I'm not saying it's definitely false, of course, but I won't be surprised to discover it ends up being more like 1-2%. (And then still hearing for years later that the DRM sapped 40% of ACO's performance.)
horkrux said:
That's why i said educated guess and i haven't thrown around "30%" and i said only the publisher who has access to both obfuscated and "regular" binary can know this for certain. On the other hand, pretending that encrypted code that is interpreted from a different instruction set (twice !! ) DOES NOT have any performance impact is simply moronic......
dr_rus
Member
There are proof in the other direction however, take Rime for example for which same outlandish claims were made and there were benchmarks which showed zero change between the Denuvo protected and later patched out versions.
I am running on a Ryzen 1700x and I have zero problems. I thought everyone in the pc master race had top of the line CPUs and 1080ti? At least thats the config all the comparison videos use.
My computer is the 1700x with a 1070, I also have the ps4 pro. So I can pretty much choose were I want to buy a game. I kind of actually like to see these types of topics because it shows the reality of PC game instead of PC master race side of things. Very few people have super computers yet every one makes the comparison to console to a top of the line PC. When in reality most people run older CPU with maybe a mid range GPU.
I am not saying that processes that run in the background should be used. But if the pc master race was real. These things wouldn't really matter.
My computer is the 1700x with a 1070, I also have the ps4 pro. So I can pretty much choose were I want to buy a game. I kind of actually like to see these types of topics because it shows the reality of PC game instead of PC master race side of things. Very few people have super computers yet every one makes the comparison to console to a top of the line PC. When in reality most people run older CPU with maybe a mid range GPU.
I am not saying that processes that run in the background should be used. But if the pc master race was real. These things wouldn't really matter.
rodrigolfp
Haptic Gamepads 4 Life
LOL. Maybe it's not real because the master race thing is just a joke to piss off some fanboys?
kruis
Exposing the sinister cartel of retailers who allow companies to pay for advertising space.
Since there is no AC:O without DRM, there is no way to verify this claim. 30-40% CPU utilization is a "tales from my ass" rumor.
Sentenza
Member
Nope.
There were previous concerns about performance hits, though. But they werent debunked. Just occasionally handwaved by official sources.
There are at least a couple of cases where Denuvos removal had a noticeable benefit on performances on the other hand.
ape2man
Member
wtf? why is this? why not switch to opengl or metal ( i forgot the name )
Locuza
Member
A DX11/OpenGL driver is incredible complex which does the majority of the memory management, checking for correctness, solving dependencies, making hazard checks, filling the command-buffers and so on.
AMD has less employees and far worse financial results, so it shouldn't be too suprising that in some cases AMD is noticeably behind Nvidia.
The OpenGL driver for Windows is pretty bad at AMD.
Metal is an Apple exclusive API for iOS und OSX.
The new explicit APIs are called DX12 and Vulkan.
But the API is choosen by the developers and implementing a good DX12/Vulkan backend is no easy job either.
Lister
Banned
Lol what a pathetic post.
rodrigolfp
Haptic Gamepads 4 Life
Which cases?
rodrigolfp
Haptic Gamepads 4 Life
Nothing about fps boost and CPU usage there. Only loading times and stuttering.
rodrigolfp
Haptic Gamepads 4 Life
What do u think was causing the stuttering ?
The thread is about CPU usage.
luca_29_bg
Member
Nope, Rime run better on my rig without denuvo, and many people on steam forum said the same.
Well, vmp basically sandbox everything, and unless there are some kind of "computer magic", sandboxed code will always runs slower than native. So yeah, this could explain the high cpu usage. But right now, it's just a word from a cracker, easily could be another case of "tales form my ass".
kirankara78
Member
High CPU usage is one thing, but even it's using more cores fully and you don't see any improvement over less cores being used, that's clearly messed up and can't be explained by an open world or ai
One more reason I've stopped gaming on PC and moved to Console.
Thank God for the X1X making the performance argument pretty much moot.
rodrigolfp
Haptic Gamepads 4 Life
Because the perfomance on XBOX will be better than on PC? Thanks for the laughs.
dr_rus
Member
No matter what people anywhere said, the benchmarks hasn't shown any difference. As in actual numbers, not words.
SmoothRunningGun
Member
So this "news" is from a pirate group? Ok.
Night Terror
Member
Well, that explains 70% CPU usage on the Main Menu I guess.
luca_29_bg
Member
Sorry, I trust more my fps counter than benchmark.
Beerman462
Member
So much wrong in one post.
This is a rumor not reality.
Pretty much any non mobile CPU is better than what's in the consoles right now, so it's a non factor. With the CPU out of the picture the GPU scaling is easy to gauge. A GPU that is 50% of the power of a 1080Ti is going to have 50% of the performance. What's worse than the PC specs is the benching games at Max! settings.
There are plenty of options when it comes to games bench-marked and reviewed. Just check out any number of videos on You Tube for low end or value builds.
The people that use PC Master Race un-ironically all seem to use it in a derogatory manner. I pretty much only hang out on GAF though.
Locuza
Member
And what was your FPS counter roughly showing before and after the patch?
30-40% performance increase?
luca_29_bg
Member
The game held the minimun 30 fps much more stable, no stuttering globally speaking and more less cpu usage. Plus in the same locations where, with the denuvo build, the cpu usage became too high with frames dropped as result, the denuvo free version kept the 30 fps more easily and with less cpu usage.
Alexander DeLarge
Member
Thanks to the gaming press not doing their jobs when it comes to actually investigating and reporting on this shit show, yeah.
What benchmarks? I haven't seen anyone except a couple YouTube videos comparing before and after the removals. The videos were (literally) a night and day performance fix. The night before it had Denuvo and it ran horribly, the following day when the only thing that changed was a Denuvo removal and it ran great for a lot of people.
For everyone spreading misinformation about how Denuvo tanks performance in most games, stop. The fact that Denuvo is a ticking time bomb for consumers is more than enough to condemn it for. The game you paid for will be unplayable by "legitimate" means the second those AWS servers go down. You will be at the mercy of the publishers to remove it