Windows 10 Game Mode Tested

[Quad Damage]

Game mode reserves 3/4 of your CPU for the game when it's the foreground application, and moves all background processes to the remaining 1/4.

If you have nothing else running on the PC like your test setup, you're basically throwing away 25% of your CPU performance, which is why you would see a reduction in performance.

If, however, you normally have media players, voip clients, web browsers and other applications open in the background when you're gaming, restricting them to using 1/4 of the CPU at most while the game is the foreground application can help improve performance.

Even if those background tasks do not total 1/4 of your CPU, removing any possibility of thread contention can help with consistency even if the average may drop a little.

 

Game Mode is especially good for games which do not fully utilize your CPU though.

If you were to play a game like Crysis - which you showed in your video but apparently did not test - that would show improvements on the quad-core CPUs when using Game Mode because the game only uses two main threads, and giving them exclusive access to a CPU core each ensures that no background processes can eat into your precious single-core performance.

 

It's not only old games which are affected by this.

Forza Horizon 3 is one of the games Microsoft uses as an example because it's so poorly optimized and the majority of the game's workload runs on a single thread.

Enabling Game Mode guarantees that main thread has exclusive access to a CPU core all to itself, instead of potentially being shared with background processes.

 

If you're trying to play a game like Deus Ex Mankind Divided - a game which requires more CPU power than a quad-core can provide - enabling Game Mode is effectively removing a core from an already CPU-starved game.

Unless, of course, your background processes total more than 1/4 of your CPU's capabilities, in which case Game Mode will limit their impact on the game's performance.

 

Since many games are only built to utilize 4 cores or up to 8 threads, I've found Game Mode to be quite useful on higher core-count CPUs like AMD's Ryzen because it will completely reserve 9 threads for the game on an R5 or 12 on an R7.

Even if you don't have much else open, there's still some CPU activity from system processes, and Game Mode ensures it has no effect on the game with these CPUs.

[captain_to_fire]

While watching the video, I was about to comment on why Linus didn't tested integrated graphics like the Intel Iris Pro of the Skull Canyon Nuc or the HD Graphics 630 of the i7-7700k but as it turns out, Game Mode won't do it any favors.

 

We can safely dismiss Windows 10 Game Mode as non functioning bloatware. 

[DKL]

Is what Quad Damage pointed out legit? I'd like to see some sources on that because I haven't heard how it actually works before.

[DildorTheDecent]

"Ashes of the Benchmark"

 

lol. That's like the only good thing about that "game"

[zMeul]

duh!

[Mira Yurizaki]

1 hour ago, Quad Damage said:

If you were to play a game like Crysis - which you showed in your video but apparently did not test - that would show improvements on the quad-core CPUs when using Game Mode because the game only uses two main threads, and giving them exclusive access to a CPU core each ensures that no background processes can eat into your precious single-core performance.

I'd like a source on this.

 

If that was really the case then anyone who has a CPU graph would show 3/4s of their CPU core's utilization spiking up, but I haven't seen that in anyone's test that I've seen.

[zMeul]

1 hour ago, Quad Damage said:

Game mode reserves 3/4 of your CPU for the game when it's the foreground application, and moves all background processes to the remaining 1/4.

If you have nothing else running on the PC like your test setup, you're basically throwing away 25% of your CPU performance, which is why you would see a reduction in performance.

no it does not, wtf are you talking about?! have any sources for those outlandish claims?

[Mira Yurizaki]

2 hours ago, Chasem121 said:

On my absolutely shitty laptop with a A6-4400 APU, it did noticeably make a RPGMaker game run better. The game was hitching every few seconds, but after I turned game mode on the hitching disappeared.

My hypothesis is that this shit is made for APUs in laptops, possibly helps keep the temps down on both the GPU cores and the CPU cores.

That's pretty much the point of this and "low-level" API. If your processing power is able to saturate the GPU, then nothing is going to make your game performance better. i.e., if you're already GPU bottlenecked, there's no point in lowering the CPU performance bar. Or at least, you shouldn't be expecting a gain in performance if you lower your CPU performance bar.

[Techicolors]

3 hours ago, Jeroen3 said:

There is a similar third party product on the market called Process Lasso.

If also features a game mode, and things it does are:

- Enable high performance power profile. (eg: no multiplier speedsteps or core parkgings)

- Put the game process in a higher scheduling priority.

 

What Windows could add here are:

- Clear more RAM by moving other stuff to the pagefile.

- Disable the Windows button.

 

In general, a "game mode" won't make any difference if all your PC is doing is gaming.

But if you PC is also downloading stuff, or even the complete worse you can do during a game, run a full heuristic virus scan.

Both above solutions can put the virus scan to the lowest priority ever, making all the resources available to the game.

 

The benchmarks performed in this video are therefore deceiving, and don't show much research into the subject. It's just a waste of time to perform those benchmarks.

Instead, try the benchmarks again, but now with prime 95 running on 75% of the cores.

i've tried process lasso once. i didn't tinker much with it, just went with the various recommended presets. after playing a heavy game for a few minutes and alt+tabbing, my pc locked up. and i alt+tabbed many times prior 

2 hours ago, Quad Damage said:

It's not only old games which are affected by this.

Forza Horizon 3 is one of the games Microsoft uses as an example because it's so poorly optimized and the majority of the game's workload runs on a single thread.

Enabling Game Mode guarantees that main thread has exclusive access to a CPU core all to itself, instead of potentially being shared with background processes.

you bring up a good point with fh3. i might test this out since the recent versions of rivatuner added overlay support for UWP games. though i have tried game mode (at least i think it was enabled) and i saw my CPU usages to be exactly the same with game mode off. 

Edited May 1, 2017 by Technicolors

[peanutsrevenge]

Initially I figured it would just be increasing the priority of the executable and setting the performance/power level to maximum.
The second of those two COULD (if done really badly) be part of the problem for worse performance on Kaby however as it's taking control away from the CPU, which would otherwise be advantageous on older CPUs, but that's a real stretch.

I think Linus nailed it, this is the return of Vista. Another good thing for Linux \o/

[Quad Damage]

9 hours ago, DKL said:

Is what Quad Damage pointed out legit? I'd like to see some sources on that because I haven't heard how it actually works before.

 

8 hours ago, M.Yurizaki said:

I'd like a source on this.

If that was really the case then anyone who has a CPU graph would show 3/4s of their CPU core's utilization spiking up, but I haven't seen that in anyone's test that I've seen.

 

8 hours ago, zMeul said:

no it does not, wtf are you talking about?! have any sources for those outlandish claims?

It's easy to test this out for yourself. Just leave a CPU monitoring program running in the background and toggle game mode.

 

For this test I disabled hardware acceleration in Firefox and played a pair of 1440p60 videos which had about 20-30% CPU usage total.

 

Alt-tab to Crysis with Game Mode on, and all other processes are moved to the first 4 threads, while Crysis gets 3/4 of the CPU to itself... even if it's not like Crysis needs that much of the CPU to be free since it's practically a single-threaded game.

 

Alt-tab away from Crysis and the restriction is removed, as Game Mode is only in effect when the game is the foreground application.

I would recommend that you avoid loading up the first 1/4 of the CPU completely like this though.

While the game itself is running smoothly, mouse aiming in the game was very laggy. I guess the interrupts always run on CPU0 or something like that.

That's not usually an issue, so long as you are not intentionally setting up a lot of background CPU usage for a demonstration like this.

I guess it might be an issue on dual-cores or slower quad-cores without hyperthreading.

[zMeul]

48 minutes ago, Quad Damage said:

It's easy to test this out for yourself. Just leave a CPU monitoring program running in the background and toggle game mode.

sorry, but no .. this isn't true

i tested this myself last month on couple of games - see one of my posts above where I linked the thread

and if need be, I'll provide additional data

[KuJoe]

12 hours ago, Quad Damage said:

If you're trying to play a game like Deus Ex Mankind Divided - a game which requires more CPU power than a quad-core can provide - enabling Game Mode is effectively removing a core from an already CPU-starved game.

Unless, of course, your background processes total more than 1/4 of your CPU's capabilities, in which case Game Mode will limit their impact on the game's performance.

I tested out Game Mode on my desktop using the Deus Ex: MD benchmark with some interesting results:

 

DX11 w/o Game Mode: 30 FPS Average, 15 Min, 46 Max

DX11 w/ Game Mode: 30 FPS Average, 28 Min, 44 Max

DX12 w/o Game Mode: 42 FPS Average, 31 Min, 53 Max

DX12 w/ Game Mode: 42 FPS Average, 33 Min, 53 Max

 

In DX11, Game Mode consistently increased the minimum FPS every run compared to runs without Game Mode. Unfortunately in DX12 it did nothing.

[SpaceGhostC2C]

On 4/30/2017 at 11:08 PM, Jeroen3 said:

There is a similar third party product on the market called Process Lasso.

There are several third party programs with equivalent functionality out there. I don't remember their names because I wouldn't know they existed if it wasn't for Microsoft (other than "Gaming Evolve" by AMD and something similar for Nvidia): around a year ago (might be inaccurate ), Microsoft sent a survey to people in the Insider Program who had ticked "gaming" as one of their uses for the PC. It included all sorts of questions about what wold improve your gaming experience, what could an OS do to improve your gaming experience, and finally whether you knew, and whether you were using, a number of third-party programs all of which promised an enhanced gaming experience, several doing exactly what Gaming Mode does now.

 

So that answers the question "why did they do it", I guess: because people were using something like this made by others. Or because enough people answered "you could hog fewer resources" in the "what would improve..." questions.

 

On 4/30/2017 at 11:11 PM, JCHelios said:

The 99.9% results to be precise.  Still really awful, though!  

In Gears of War, when Linus mentioned that "only the high-end system saw a big difference, but who cares about 150 vs 140 average FPS", he forgot to mention the difference was in favor of "Game Mode OFF"

Although nothing as drastic as the minimum results, even for averages results were often against Game Mode, (just most of the times within margin of error).

 

I also think people should stop calling minimum frames a "consistency test", as it's almost an extreme value test, there are many simple statistics that better capture the idea of "what to typically expect, regardless of the odd high or low value here and there" (like mode, median, variance, etc. Or the centered interval that encompasses X% of the values). But I digress, it's not important here  

[TheLaserCucumber]

I decided to test it myself on my HTPC (specs in signature) and I was kinda surprised.

I did 2 test runs in GTA V. The first with game mode toggled off, and the second with it switched on.

 

The first run was interesting. The min was 9.2, the max was 86 and the avg was 55.4FPS.

But there was a lot of glitching. Textures that were not loaded correctly, and parts of the world missing.

I was surpised by this as nothing of this happens in the real game (Story mode or online!).

 

Things changed when I got to the second run, with Game Mode turned on.

The min was 19.6, the max 66.4 and the avg 52.8FPS.

Seemingly the game mode reduced the max and avg FPS, but boosted the mins in comparison to the first run.

But surprisingly, the texture and world glitching wasn't happening anymore.

 

Conclusion: I think the game mode can help older systems, with kinda weak CPUs.

I will test some more games, and post them here, if someone is interested.

 

[JohnVHSTapes]

It really helps on systems like mine with older hardware where windows on it's own uses lots of resources when I'm playing games. I used to have performance problems where some days the game (BF1) would run fine at high settings with 60fps. Then other days where BF1 won't achieve 20fps on the lowest settings. I'd check task manager to see whats going on and windows update would be using 30% of my CPU on it's own. With this game mode enabled I have been seeing very consistent performance over time now. I don't have those days where the game is almost unplayable. So game mode for people like me is very useful and I'm very glad they added the feature.

[DKL]

Interesting. So Game Mode helps the game be able to RUN if your CPU is a real bottleneck to the point of being unplayable or almost unplayable. If your CPU is running the game at 60+FPS average, Game Mode likely has no use to you.

 

Game Mode is NOT something to BOOST performance but rather to STABILIZE performance. 

[XenosTech]

Tested this with my gf's laptop last weekend... Game more is meant for weaker systems i.e systems running on iGPU's or APU. She would get 60 fps in LoL but when she gets near any team fights she would end up in the 20's and below, once that was done back up to 60 like normal... Did a force update on her laptop and ran another match with game mode on and voila her max fps stayed about the same but the dips she got in a term fight weren't as massive as before, now a team fight had about 38-45 fps up from 20 and under.

So to pair a budget cpu with a decent discrete gpu shouldn't see any significant or noticeable performance from game mode being on.

[Brisingr_Phoenix]

If you are an affluent high school student then that would make sense as a budget build "for the lolz"

If you are me when I was making my first build then it would look more like: 8gb ddr3, fx 4200, and a gt 630 running under a 350 watt psu