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

Denuvo Performance Cost & FPS Loss Tested

W1zzard

Administrator
Staff member
Joined
May 14, 2004
Messages
18,827 (3.47/day)
Likes
21,285
Processor Core i7-4790K
Memory 16 GB
Video Card(s) GTX 1080
Display(s) 30" 2560x1600 + 19" 1280x1024
Software Windows 7
#1
We take a detailed look at the performance cost of the Denuvo copy protection system and how it impacts FPS. We tested over 70 scenarios: varying game resolution, varying CPU frequency, different CPU core and thread counts, and graphics hardware from AMD vs. NVIDIA.

Show full article
 
Last edited by a moderator:
Joined
Jan 27, 2019
Messages
31 (0.53/day)
Likes
28
#2
I am not sure what to believe but I will just follow my rule of thumb and believe it when I see it.
 
Last edited by a moderator:

newtekie1

Semi-Retired Folder
Joined
Nov 22, 2005
Messages
26,009 (5.34/day)
Likes
12,442
Location
Indiana, USA
Processor Intel Core i7 8700K@4.8GHz(Quick and dirty)
Motherboard AsRock Z370 Taichi
Cooling Corsair H110i GTX w/ Noctua NF-A14 Fans
Memory 32GB Corsair DDR4-3000
Video Card(s) ASUS Strix GTX 1080Ti
Storage 500GB Crucial MX500 + 2TB Seagate Solid State Hybrid Drive with 480GB MX200 SSD Cache
Display(s) QNIX QX2710 1440p@120Hz
Case Fractal Design Define S
Audio Device(s) Onboard is good enough for me
Power Supply Corsair HX850
Software Windows 10 Pro x64
#3
So, basically, this proves that when implemented correctly Denuvo will have no noticeable performance impact.
 
Joined
Sep 17, 2014
Messages
8,164 (4.94/day)
Likes
7,202
Location
Duiven, Netherlands
Processor i7 8700k 4.7Ghz @ 1.26v
Motherboard AsRock Fatal1ty K6 Z370
Cooling beQuiet! Dark Rock Pro 3
Memory 16GB Corsair Vengeance LPX 3200/C16
Video Card(s) MSI GTX 1080 Gaming X @ 2100/5500
Storage Samsung 850 EVO 1TB + Samsung 830 256GB + Crucial BX100 250GB + Toshiba 1TB HDD
Display(s) Eizo Foris FG2421
Case Fractal Design Define C TG
Power Supply EVGA G2 750w
Mouse Logitech G502 Protheus Spectrum
Keyboard Sharkoon MK80 (Brown)
Software W10 x64
#4
when implemented correctly
This being the key part of that phrase. We've seen many implementations all different, and there is absolutely no guarantee that future versions of Denuvo will allow the same kind of implementation as the battle between cracking scene and them goes on.

Also, DMC5 is a super light game, that is linear and extremely CPU friendly. When resources are not saturated of course you'll barely notice it... I would love for this test to be revisited in games like AC: Odyssey or Deus Ex Mankind Divided, games that show heavy tanking of FPS due to CPU loads.

Nonetheless a great article, very informative.
 

W1zzard

Administrator
Staff member
Joined
May 14, 2004
Messages
18,827 (3.47/day)
Likes
21,285
Processor Core i7-4790K
Memory 16 GB
Video Card(s) GTX 1080
Display(s) 30" 2560x1600 + 19" 1280x1024
Software Windows 7
#5
This being the key part of that phrase. We've seen many implementations all different, and there is absolutely no guarantee that future versions of Denuvo will allow the same kind of implementation as the battle between cracking scene and them goes on.
I'm not seeing any reason why Denuvo should be placed in a performance critical code path, and I'm having a hard time imagining that any developer would be so stupid to do that

When resources are not saturated
They are saturated, either GPU or CPU, depending on the test (I get what you're saying though)

like AC: Odyssey or Deus Ex Mankind Divided
Impossible unfortunately unless they release a Denuvo-free EXE that has no other code changes. That's what makes the case of DmC 5 so interesting
 
Joined
Mar 13, 2014
Messages
3,481 (1.89/day)
Likes
4,313
Processor i7 7700k
Motherboard MSI Z270 SLI Plus
Cooling CM Hyper 212 EVO
Memory 2 x 8 GB Corsair Vengeance
Video Card(s) MSI GTX 980 Ti GAMING
Storage Samsung 850 EVO 250 GB and WD Black 4TB
Display(s) ASUS 27 inch 1440p PLS PB278Q
Case Corsair Obsidian 750D Airflow Edition
Audio Device(s) Onboard
Power Supply EVGA SuperNova 850 W Gold
Mouse Logitech G502
Keyboard Logitech G105
Software Windows 10
#7
From the article:

"Another gotcha is what happens with the game when the activation servers are down or gets turned off a decade from now. You would have paid for the game but might not be able to play it. "

I don't like DRM in general and that's why I buy from GOG whenever possible but that is my biggest concern with Denuvo. If I enjoy a game more than average then I replay it every few years. I even go back as far as the late 90's sometimes. I don't want that to be taken away from me if Denuvo should disappear in the future.

A lot of Developers remove Denuvo after the game has been cracked but not all do.
 
Joined
Aug 27, 2015
Messages
443 (0.34/day)
Likes
290
Location
In the middle of nowhere
System Name Scrapped Parts, Unite !
Processor i5-2500 @ 3.3 Ghz//Q9300 @ 2.5 Ghz
Motherboard GA-H67MA-USB3-B3// P35 Neo2-FR
Cooling Stock
Memory 10 GB DDR3 1333 // 3 GB DDR2
Video Card(s) Colorful iGame GeForce GTX1060 Vulcan U 6G // ASUS ENGT 220 1GD3
Storage Hitachi 500 GB, Sony 1TB, KINGSTON 400A 120GB // Samsung 160 GB
Display(s) HP 2009f
Case Xigmatek Asgard Pro // Cooler Master Centurion 5
Audio Device(s) Onboard
Power Supply OCZ ModXStream Pro 500 W
Software Windows 10 x64
Benchmark Scores Minesweeper 30fps, Tetris 40 fps, with overheated CPU and GPU
#8
Another gotcha is what happens with the game when the activation servers are down or gets turned off a decade from now. You would have paid for the game but might not be able to play it.
2000 = you PAID the game, you OWN it
2019 = you PAID the game, you RENT it
 
Joined
Feb 19, 2009
Messages
890 (0.24/day)
Likes
225
Location
I live in Norway, in the province Buskerud.
System Name 2 sys spec seperated by "|"
Processor R7 1700 @ 3.9| Xeon 2680v2
Motherboard Asrock B350M Pro 4 | HP Z620
Cooling Water1x360,1x240 | Air
Memory 64gb G.skill SniperX 3200mhz CL16 | 128 gb
Video Card(s) EK-FC - RX Vega 64 | GTX970
Storage 960 Evo 500gb, 860 evo 1tb, Intel 660P 1tb PM871 256gb | various intel ssd's.
Display(s) AOC 28" 4K something + 1440p 144hz something.
Case Phanteks EvolvX M-Atx
Power Supply Corsair RM850
Mouse G402
Keyboard TT Meka G1
#9
I feel they can have correctly implemented denuvo for a max of 3 years then remove it.
 
Joined
Oct 1, 2006
Messages
4,230 (0.93/day)
Likes
634
Location
Hong Kong
Processor Ryzen Threadripper 1950X
Motherboard X399 Fatal1ty Professional Gaming
Memory Team NightHawk 3600 4x8GB
Video Card(s) Zotac RTX 2080 Ti
Storage Samsung 960 Evo 512GB 3-way Raid 0
Display(s) Samsung C27HG70
Case Thermaltake Level 20 XT
Audio Device(s) Xonar Essence STX II
Power Supply FSP Aurum PT 1200W
Software Windows 10 Pro 64-bit
#10
So, basically, this proves that when implemented correctly Denuvo will have no noticeable performance impact.
Although the difference is small in this case, it is indeed a consistent and measureable difference.
So the point still stands that Denuvo hurts perfromance for paying customers.
 
Joined
Sep 17, 2014
Messages
8,164 (4.94/day)
Likes
7,202
Location
Duiven, Netherlands
Processor i7 8700k 4.7Ghz @ 1.26v
Motherboard AsRock Fatal1ty K6 Z370
Cooling beQuiet! Dark Rock Pro 3
Memory 16GB Corsair Vengeance LPX 3200/C16
Video Card(s) MSI GTX 1080 Gaming X @ 2100/5500
Storage Samsung 850 EVO 1TB + Samsung 830 256GB + Crucial BX100 250GB + Toshiba 1TB HDD
Display(s) Eizo Foris FG2421
Case Fractal Design Define C TG
Power Supply EVGA G2 750w
Mouse Logitech G502 Protheus Spectrum
Keyboard Sharkoon MK80 (Brown)
Software W10 x64
#11
I'm not seeing any reason why Denuvo should be placed in a performance critical code path, and I'm having a hard time imagining that any developer would be so stupid to do that
I'm not talking about what developers would do, because of course they would not willingly cripple performance. The reality though is that developers don't always have the final say. The publisher can decide version X.X of Denuvo needs to be used and Denuvo can release versions that define how it needs to be implemented to make sure it actually works. We've already seen major differences between different versions of Denuvo (crackable vs. very hard to crack) and Denuvo's whole business relies on the effectiveness of the DRM.

I think there is no telling or guarantee that Denuvo will not be placed anywhere.
 
Joined
May 22, 2015
Messages
5,380 (3.83/day)
Likes
2,354
Processor Intel i5-6600k
Motherboard ASRock Z170 Extreme7+
Cooling Arctic Cooling Freezer i11
Memory 2x8GB DDR4 2400 G.Skill
Video Card(s) EVGA GTX 1060 SC
Storage 128 and 256GB OCZ Vertex4, 500GB Samsung 850 EVO, 1TB Crucial MX300 and 3TB Seagate
Display(s) HP ZR24w
Case Chieftec BX01
Power Supply Seasonic 620W M12
Mouse Logitech G502 Proteus Core
Keyboard G.Skill KM780R
#12
While I have posted in the past we had no hard evidence Denuvo was impacting performance, my main concern is layered DRM. Do I need Denuvo running under SteamWorks possibly running under yet another DRM layer? Do all parties play nice with each other?
I just stuck with GOG and never cared much in the end.
 

newtekie1

Semi-Retired Folder
Joined
Nov 22, 2005
Messages
26,009 (5.34/day)
Likes
12,442
Location
Indiana, USA
Processor Intel Core i7 8700K@4.8GHz(Quick and dirty)
Motherboard AsRock Z370 Taichi
Cooling Corsair H110i GTX w/ Noctua NF-A14 Fans
Memory 32GB Corsair DDR4-3000
Video Card(s) ASUS Strix GTX 1080Ti
Storage 500GB Crucial MX500 + 2TB Seagate Solid State Hybrid Drive with 480GB MX200 SSD Cache
Display(s) QNIX QX2710 1440p@120Hz
Case Fractal Design Define S
Audio Device(s) Onboard is good enough for me
Power Supply Corsair HX850
Software Windows 10 Pro x64
#13
Although the difference is small in this case, it is indeed a consistent and measureable difference.
So the point still stands that Denuvo hurts perfromance for paying customers.
There was never an argument that it didn't. The point is a measurable difference is not a noticeable difference. If the paying customer doesn't notice the difference, their experience playing the game is not changed, and there is no problem.
 

W1zzard

Administrator
Staff member
Joined
May 14, 2004
Messages
18,827 (3.47/day)
Likes
21,285
Processor Core i7-4790K
Memory 16 GB
Video Card(s) GTX 1080
Display(s) 30" 2560x1600 + 19" 1280x1024
Software Windows 7
#14
The publisher can decide version X.X of Denuvo needs to be used and Denuvo can release versions that define how it needs to be implemented to make sure it actually works
But it's the developer who puts in the Denuvo calls, the publisher can not do that as you need to recompile from source code. Just to clarify for other readers, if Denuvo releases an update, older games with older Denuvo versions won't be affected "automagically", this is not Windows Update

We've already seen major differences between different versions of Denuvo (crackable vs. very hard to crack) and Denuvo's whole business relies on the effectiveness of the DRM.
I think there is no telling or guarantee that Denuvo will not be placed anywhere.
The thing is that it makes no difference whether you place the Denuvo calls in the render thread or in any other code that gets regularly executed, only one slows down rendering, the other does not. From a cracking perspective these are 100% identical
 
Joined
May 31, 2017
Messages
99 (0.15/day)
Likes
33
Processor i7-4790k 4.6Ghz
Motherboard Gigabyte Z97 Sniper.G1
Cooling Thermalright HR-02
Memory G.Skill 2400Mhz @ C10
Video Card(s) Gigabyte RX 580 Aorus
Storage Seagate Barracuda 2TB
Display(s) AOC G2260VWQ6
Audio Device(s) SB Recon 3Di
Power Supply Super Flower Golden Green HX 550w
Mouse Logitech G403
Keyboard Office Keyboard
Software W10 build 17763
#15
now compare cpu usage, this benchmark is useless

i like to use my computer for other taks when gaming
 
Joined
Jun 28, 2015
Messages
338 (0.25/day)
Likes
134
#16
With game engines finally upgrading to be able to use more threads & CPU cores DRM impact should be less visible on multicore processors.

But DRM all comes down to renting games now for a period of time, not ownership in the traditional sense anymore.
 
Joined
May 22, 2015
Messages
5,380 (3.83/day)
Likes
2,354
Processor Intel i5-6600k
Motherboard ASRock Z170 Extreme7+
Cooling Arctic Cooling Freezer i11
Memory 2x8GB DDR4 2400 G.Skill
Video Card(s) EVGA GTX 1060 SC
Storage 128 and 256GB OCZ Vertex4, 500GB Samsung 850 EVO, 1TB Crucial MX300 and 3TB Seagate
Display(s) HP ZR24w
Case Chieftec BX01
Power Supply Seasonic 620W M12
Mouse Logitech G502 Proteus Core
Keyboard G.Skill KM780R
#17
With game engines finally upgrading to be able to use more threads & CPU cores DRM impact should be less visible on multicore processors.

But DRM all comes down to renting games now for a period of time, not ownership in the traditional sense anymore.
Ownership has always been governed by the EULA which stipulates you only license the right to play the game. But it's true on PC, in the absence of DRM, you pretty much owned what you paid for. But that has never been true on consoles.
 
Joined
Sep 17, 2014
Messages
8,164 (4.94/day)
Likes
7,202
Location
Duiven, Netherlands
Processor i7 8700k 4.7Ghz @ 1.26v
Motherboard AsRock Fatal1ty K6 Z370
Cooling beQuiet! Dark Rock Pro 3
Memory 16GB Corsair Vengeance LPX 3200/C16
Video Card(s) MSI GTX 1080 Gaming X @ 2100/5500
Storage Samsung 850 EVO 1TB + Samsung 830 256GB + Crucial BX100 250GB + Toshiba 1TB HDD
Display(s) Eizo Foris FG2421
Case Fractal Design Define C TG
Power Supply EVGA G2 750w
Mouse Logitech G502 Protheus Spectrum
Keyboard Sharkoon MK80 (Brown)
Software W10 x64
#18
But it's the developer who puts in the Denuvo calls, the publisher can not do that as you need to recompile from source code. Just to clarify for other readers, if Denuvo releases an update, older games with older Denuvo versions won't be affected "automagically", this is not Windows Update


The thing is that it makes no difference whether you place the Denuvo calls in the render thread or in any other code that gets regularly executed, only one slows down rendering, the other does not. From a cracking perspective these are 100% identical
Alright, I see.

Is there anything you can say about stuff like VMware VMProtect (thanks W1z) that was used on top of Denuvo for example on AC:Odyssey?
 
Last edited:

W1zzard

Administrator
Staff member
Joined
May 14, 2004
Messages
18,827 (3.47/day)
Likes
21,285
Processor Core i7-4790K
Memory 16 GB
Video Card(s) GTX 1080
Display(s) 30" 2560x1600 + 19" 1280x1024
Software Windows 7
#20
Is there anything you can say about stuff like VMware that was used on top of Denuvo for example on AC:Odyssey?
VMProtect? http://vmpsoft.com/support/user-manual/introduction/what-is-vmprotect/

"The method of application code mutation applied in VMProtect is based on obfuscation — a process that adds to the application code various excessive, “garbage” commands, “dead” parts of the code, random conditional jumps. It also mutates original commands and transfers execution of certain operations to the stack."
All this code will use CPU time which shouldn't be a problem for "normal" applications. For games this becomes relevant though, unless you exclude performance-sensitive sections of the code from VMProtect (which is possible, but may be undesired using a naive view on things).

Combining this with Denuvo though will multiply (not combine) the performance impact. As we've seen from the EXE size, Denuvo adds A LOT of code to the binary, VMProtect amplifies that by adding a lot of obfuscation code around the larger code size.

I can understand the desire for using both though. Denuvo doesn't scramble the executable, so you can "just" load it into IDA Pro and start understanding how the program works (understanding Denuvo code is VERY complicated). Putting VMProtect on top of that makes things much much more difficult, but apparently not impossible, considering these games have been cracked in the meantime.
 
Joined
Jun 28, 2015
Messages
338 (0.25/day)
Likes
134
#21
Ownership has always been governed by the EULA which stipulates you only license the right to play the game. But it's true on PC, in the absence of DRM, you pretty much owned what you paid for. But that has never been true on consoles.
Indeed.
 
Joined
Jan 31, 2010
Messages
2,998 (0.90/day)
Likes
1,106
System Name Cumquat 2000
Processor AMD FX8320 @ 4.0GHz NB/HT @ 2.6GHz
Motherboard Asus Crosshair V Formula
Cooling ThermoLab Baram 2x 120mm Gentle Typhoons 1850rpm
Memory 8GB (2x4GB) Mushkin Blackline DDR3 2400MHz 11/13/13/34/1T 1.65V
Video Card(s) Sapphire Nitro+ RX580 8GB OC 1342/2000 (Core/Mem) BIOS Modded
Storage 1X Samsung 860 EVO 500GB SATA AHCI 1x Sandisk 128GB SSD, 1xWD 2TB red, 1xWD 1TB black, 1xWD 2TB nas
Display(s) ViewSonic VX2457 MHD 75Hz FreeSync 1920x1080p
Case SilverStone Raven RV02B-W + 3x SS AP181 + USB3.0 upgrade
Audio Device(s) Creative SB X-Fi Titanium Fatal1ty Pro PCIe + Pioneer VSX-D457 amp Technics Speaker set
Power Supply SilverStone ST75F-P Modular 750W
Mouse Logitech G502 Proteus Core
Keyboard Logitech G105
Software Windows 10 X64 PRO (build 1803)
Benchmark Scores it sucks even less much now ;)
#22
So it shows no noticeable (or determinable) difference when playing a game using a high end intel CPU big deal not everyone can afford a new cpu everytime one comes out would be nice to see what happens with say an FX8350 or i7 3770 is used at stock speeds with the same set of GPU's used here
 
Joined
Sep 17, 2014
Messages
8,164 (4.94/day)
Likes
7,202
Location
Duiven, Netherlands
Processor i7 8700k 4.7Ghz @ 1.26v
Motherboard AsRock Fatal1ty K6 Z370
Cooling beQuiet! Dark Rock Pro 3
Memory 16GB Corsair Vengeance LPX 3200/C16
Video Card(s) MSI GTX 1080 Gaming X @ 2100/5500
Storage Samsung 850 EVO 1TB + Samsung 830 256GB + Crucial BX100 250GB + Toshiba 1TB HDD
Display(s) Eizo Foris FG2421
Case Fractal Design Define C TG
Power Supply EVGA G2 750w
Mouse Logitech G502 Protheus Spectrum
Keyboard Sharkoon MK80 (Brown)
Software W10 x64
#23
So it shows no noticeable (or determinable) difference when playing a game using a high end intel CPU big deal not everyone can afford a new cpu everytime one comes out would be nice to see what happens with say an FX8350 or i7 3770 is used at stock speeds with the same set of GPU's used here
Do you even read, the test contains an 8700K that was restricted to almost every core count / speed configuration you can find in a CPU product stack.

The only possible gripe you can have with this test is that only one game, and one that isn't CPU heavy, was used, but if the implementation is similar, the perf loss will also be similar. 3-4% of actual FPS loss, and that the same across all CPU configurations in the test - so even on a 2c/2t CPU.

By the way, if you use an FX or i7 3770 with your new 2080ti you need a lesson on overall system balance.
 
Joined
Jan 31, 2010
Messages
2,998 (0.90/day)
Likes
1,106
System Name Cumquat 2000
Processor AMD FX8320 @ 4.0GHz NB/HT @ 2.6GHz
Motherboard Asus Crosshair V Formula
Cooling ThermoLab Baram 2x 120mm Gentle Typhoons 1850rpm
Memory 8GB (2x4GB) Mushkin Blackline DDR3 2400MHz 11/13/13/34/1T 1.65V
Video Card(s) Sapphire Nitro+ RX580 8GB OC 1342/2000 (Core/Mem) BIOS Modded
Storage 1X Samsung 860 EVO 500GB SATA AHCI 1x Sandisk 128GB SSD, 1xWD 2TB red, 1xWD 1TB black, 1xWD 2TB nas
Display(s) ViewSonic VX2457 MHD 75Hz FreeSync 1920x1080p
Case SilverStone Raven RV02B-W + 3x SS AP181 + USB3.0 upgrade
Audio Device(s) Creative SB X-Fi Titanium Fatal1ty Pro PCIe + Pioneer VSX-D457 amp Technics Speaker set
Power Supply SilverStone ST75F-P Modular 750W
Mouse Logitech G502 Proteus Core
Keyboard Logitech G105
Software Windows 10 X64 PRO (build 1803)
Benchmark Scores it sucks even less much now ;)
#24
By the way, if you use an FX or i7 3770 with your new 2080ti you need a lesson on overall system balance.
If I could afford the 2000 bucks odd for a 2080ti I wouldn't still be using an FX CPU but point taken about the perf
 
Joined
May 24, 2013
Messages
12 (0.01/day)
Likes
4
Processor i3570k @4.5
Motherboard Asus Maximus Gene V
Video Card(s) GTX670 SLI
Case Fractal Design Define Mini
Mouse Razer DeathAdder Chroma
Keyboard Steelseries Apex
#25
now compare cpu usage, this benchmark is useless

i like to use my computer for other taks when gaming
I don't agree that it's useless, but CPU and RAM usage would be nice as some causal gamers usually have other tasks running.
 
Top