Windows 11 vs 10 – Windows 11 up to 17% SLOWER?!

STOP EVERYTHING! Windows 11 can reduce performance in your laptop,  here’s everything that you need to know about before you consider upgrading. You might have seen some reports like this one from PCGamer that Windows 11  “can tank PC gaming performance by around 25%”. In this article,  they’re saying this is due to a feature called VBS. Without getting too technical, VBS, or Virtualization Based Security is  a suite of features in Windows that helps improve security.

The setting in particular that seems to cause performance issues for games is Core Isolation,  or memory integrity. Fortunately this is simple to change on or off in Windows,  it just needs a reboot between each change. Microsoft’s documentation notes that in most cases,  memory integrity is on by default in Windows 11, and can be turned on for Windows 10. So this isn’t really anything new then. It existed in Windows 10,  it’s just that it wasn’t on by default.

Now on my test laptop I did an in-place upgrade from  Windows 10 to Windows 11, and it remembered all of my settings including leaving core isolation off.

However, I did also try a fresh installation with the Windows 11 ISO that I downloaded from  Microsoft, and when I installed that on this same laptop core isolation was on by default. This might also mean that when you buy a brand new  Windows 11 laptop it’ll probably have that enabled by default.  I haven’t yet been sent any laptops that came with Windows 11 out of the box so I can’t verify  that just yet. But it does seem pretty likely if it was the default with a fresh install.

So if memory integrity / core isolation / VBS is probably going to be on by default,  what sort of a performance difference does it make? All of my testing has been done with the Lenovo Legion 5i gaming laptop. Mine has a 6 core Intel  i5-11400H processor and Nvidia RTX 3060 graphics. This laptop was sent to me with Windows 10 and was  updated with the Windows 11 installation assistant after it officially launched on October 5th. I felt the in-place upgrade would better represent how most of you are probably going to upgrade your  machines.

I’m sure a lot of people are going to do a fresh installation with the ISO, but I’m also  sure a lot of people are probably just going to do an in-place upgrade of their Windows 10 machine.

Now all games were tested with the same Nvidia driver, version 472.12,  which according to Nvidia has official support for both Windows 10 and Windows 11. So with all of that in mind, let’s find out what the performance differences are between  Windows 10 and Windows 11 in 12 different games, then after the games we’ll look  at things like battery life, temperature differences and content creator workloads. Let’s start out with Microsoft Flight Simulator, as it has some interesting results.

  Now give me a chance to explain this rainbow mess. The red bars are Windows 10,  purple bars are Windows 11 with VBS off, and green bars are Windows 11 but with VBS turned on. I’ve  also tested all available setting presets with lowest down the bottom and highest up the top.  This was done because generally higher setting presets are more GPU bound,  and I was curious how this affected results. Regardless of setting preset,  we can see that Windows 10 in the red bars had the highest result, though Windows 11 with VBS  also off in the purple bars was extremely close.

Windows 11 with VBS enabled in the  green bars had the biggest dip, so this security feature is definitely taking off some FPS here. That’s not always the case though, well it is, but the degree to which performance is lost seems  to depend on the game and setting preset. Take Control for instance, with all settings maxed  out we’re seeing pretty much the same result regardless of operating system version and  whether or not VBS is on or off. They’re also fairly similar with medium settings,  but then at low we can see the average FPS starts to deviate, with Windows 10 giving us the best  result, Windows 11 with VBS off just behind it, then Windows 11 and VBS on was the lowest.  This is a fairly GPU heavy game, so perhaps VBS is making a bigger difference at lower settings where  we’re less GPU bound.

I’ve also tested Control but with both ray tracing set to high and DLSS  enabled. This would be fairly GPU dependent now, even at low settings, which is probably why the  differences are like 1-2 FPS or so regardless of setting level now, so nothing major at all.

This is why I’ve tested CS:GO, as the game is heavily dependent on the processor,  and sure enough this test saw some of the biggest differences out of all 12 games tested. At max  settings the Windows 10 and 11 results with VBS off were basically the same in terms of  average frame rate, however interestingly the 1% low from Windows 11 had quite a big lead,  and this continued at lower setting presets too. With VBS on though, the 1% low from Windows 11  was about the same as Windows 10, however the average frame rate was negatively affected.

  With lowest settings for instance, VBS on is about 15% slower than VBS off in Windows 11. Cyberpunk 2077 had some very interesting results. Between low and ultra setting presets, in general  Windows 10 and 11 were quite close together, just a 1 FPS difference or less between them, margin  of error stuff, and although Windows 11 with VBS on in the green bars was lower, it’s probably not  enough of a difference in this game that you’d actually notice this when playing the game,  but performance is still lower.

What I thought was interesting were the ray tracing presets,  which I’ve also tested with DLSS set to quality. The results are kind of all over the place,  and much different compared to when I did this with Control earlier.

Windows 11 with VBS on is  definitely much worse here, while with it off it was a fair bit ahead of the Windows 10 results. Watch Dogs Legion also had some odd results at ultra settings,  to the point where I’m wondering if the Windows 10 result should have been retested.  Unfortunately I didn’t pick this above average change up until I’d already upgraded to Windows  11 though. Ultra settings aside, like most other games there’s still an obvious difference between  VBS on and off which can be seen by comparing the green and purple bars at each setting level.  In general though Windows 10 in the red bars was ahead of Windows 11,  not counting ultra settings which as mentioned might not be accurate.

Red Dead Redemption 2 on the other hand had some fairly minor differences. Basically no  change between the three results at the highest ultra setting preset, then at lower levels we’re  looking at a 2-3 FPS loss with Windows 11 and VBS enabled in green bars which was lower than  the other results. Not much of a difference, but it’s still a measurable decrease in performance. Assassin’s Creed Valhalla had small differences too.

In general Windows  10 was ahead of Windows 11 at all but max settings, but it’s like a 1-2 FPS difference,  so really margin of error stuff.

 Similar deal with Windows 11 with VBS  on and off, just a 1-2 FPS difference, nothing serious in this one either. Fortnite was tested with the exact same replay in all three instances.  The differences were very small at max settings, though the gap seems to get wider as we step  down to lower setting presets. Generally speaking Windows 11 with VBS enabled was the worst result,  and Windows 10 was always performing better than Windows 11, even with VBS disabled. Rainbow Six Siege was also hitting super high FPS regardless of setting level,  but again Windows 11 with VBS enabled in the green bars was consistently  lower compared to the other results, but yeah at above 200 FPS for even the 1% low  at max settings are you really going to be able to tell the difference?

Probably not. I’ve also got the results from three older games here too,  I’ll just skip through them so we can get into the fun stuff,  but feel free to pause the video if you want a closer look at any of these results.

These are the differences between all 12 games at the highest setting preset when we’re comparing  Windows 10 and 11 both with VBS off, so this was the default after my in place upgrade as I came  from 10 it kept the original settings of core isolation being off. We’re looking at a 0.1%  difference on average over the 12 games with only extremely minor differences one way or the other,  honestly pretty much all within margin of error if we’re being realistic.

A 3% difference in the  best case isn’t really going to be something you’re likely to notice while playing games.  I also want to note I used the 2nd highest preset for Watch Dogs here,  as mentioned I’m not certain about the ultra results and couldn’t retest.

If we instead compare the differences with all settings at minimum the newer Windows 11 was now  behind in most cases, with the biggest outlier being Fortnite down the bottom. The same laptop  with Windows 10 was just performing better in most of these games at minimum settings,  though again in most cases it’s not going to be a noticeable difference,  so I think we can say going from Windows 10 to 11 in itself doesn’t really  affect gaming performance, of course assuming Core Isolation is disabled. Now here’s where things get interesting.

Things change quite a bit with core isolation enabled  in Windows 11, and don’t forget this was the default when I did a fresh Windows 11 install.  On average, Windows 11 with core isolation on is now almost 8% slower compared to Windows 10,  which by default has core isolation off.

I’d say the games that are generally more CPU heavy  towards the bottom of the graph are seeing the largest differences. In the case of Microsoft  Flight Simulator down the bottom, if you recall it was around 81 FPS with Windows 11 and VBS on,  while Windows 10 was 98 FPS, so a fair difference in that case. If we instead compare Windows 11 against itself with VBS on and off, we can see that VBS on  is around 6% slower with all setting presets on lowest.

If we instead look at highest setting  presets then VBS on is about 3 and a half percent slower, so yeah there’s definitely  a larger difference with this security option in games at lower settings. The difference is  less significant at higher setting presets, presumably because we’re more GPU bound. So in general there’s only minor differences between Windows 10 and Windows 11,  all things being equal with core isolation off at least. With VBS / core isolation  enabled there was more of a noticeable dip.

Honestly I’d probably expect a  similar performance dip in Windows 10 if we manually enabled core isolation as well.

  But the issue is that Windows 10 didn’t have core isolation by default while Windows 11 seems to. It does depend entirely on the game and setting preset though. In most games it’s probably not  going to make that big of a difference, but if you are after max performance at  the expense of security then you might want to consider leaving core isolation disabled.

So then what about performance outside of games? I’ve got some Cinebench results to get an idea of if processor  performance is affected.

Now Cinebench scores generally don’t care too much about memory,  and given we’re changing a setting that deals with memory, that being memory integrity,  it’s not too surprising that we’re not seeing any significant difference.  Sure technically the VBS on result was the lowest, but at just 43 points below the Windows  10 result it means Windows 10 is only scoring 0.4% higher, well within the margin of error.

Adobe Premiere was tested with the Puget Systems benchmark.  Windows 10 had the best score, though while Windows 11 with VBS off was slightly lower,  this is well within margin of error for this test, however VBS on was notably lower comparatively.

It’s a similar deal in Adobe Photoshop,  pretty much the same scores between Windows 10 and 11 with VBS off,  while VBS on in Windows 11 was lower, and again it’s pretty similar in DaVinci Resolve as well. Puget Systems, the guys who make these benchmarks, came to a similar conclusion too,  noting that the Adobe applications don’t yet have official Windows 11 support,  so due to this alone they recommend holding off on upgrading at this time as  this could result in other instability issues, not just lower performance.

I’ve also compared temperatures while under heavy combined CPU plus GPU stress test.  There wasn’t really a noteworthy difference, one degree or so is margin of error stuff. Likewise the clock speeds that were being hit were also pretty much the  same regardless of operating system, and this is because for the most part  the power levels the CPU and GPU were running at were quite similar too.

I wasn’t expecting to see any temperature differences because things like CPU and GPU  power limits are generally defined by the OEM. That type of thing just doesn’t change with an  operating system upgrade, and as we’ve seen that does indeed appear to be the case here.

I wasn’t seeing a practical difference in battery life. Windows 11 was lasting  3 minutes longer but again like a lot of the tests,  that’s honestly within margin of error given both are around 5 and a half hours. According to Microsoft, Windows 11 does apparently introduce some battery improvements for laptops,  and while I didn’t happen to see any differences in my video playback test on this laptop,  Matthew Moniz has tested both an Intel and AMD laptop in the PCMark  battery test and found that Windows 11 was performing a fair bit worse.

Now I’ve only tested one configuration of laptop here. Considering that there seems  to be a larger performance difference in games at lower setting levels,  it might make sense that laptops that have lower specs could also see similar differences.  Say if you have a lower tier GTX 1650 graphics or something, perhaps in many games you won’t  be able to run at those high setting presets. While I haven’t explicitly tested a laptop with  lower specs than this it would make sense that with lower spec hardware you have to  run at lower settings, and yeah as we did see VBS on did make more of a difference there.

All of my testing has been done with an Intel based system.

AMD have noted that their Ryzen  processors may have reduced performance in certain applications when running Windows 11,  and they explicitly state that 10-15% performance loss is possible in esports games. This page also notes that both problems will be resolved later this month with updates,  so as this is going to be fixed soon anyway, I haven’t spent multiple days  doing all the retesting on a Ryzen laptop as well. Basically just be aware that if  you do have an AMD based laptop until those updates come out you might see  even worse performance compared to what I’ve shown here with Windows 11 on Intel. Now something that Windows 11 will actually be required for is DirectStorage. With an NVMe  SSD Windows 11 will be able to load game assets straight to the GPU without loading up the CPU,  so basically faster game load times and more performance as the CPU is more available.

As far as I’m aware, this isn’t yet implemented in any current games, and support will also depend  on the game. But it absolutely sounds like a cool feature that I can’t wait to test out, and Windows  11 also has some other improvements like auto HDR which should help improve the HDR experience in  gaming. So yeah, Windows 11 certainly sounds promising with those upcoming features but  right now a few days after the official launch I’d probably hold off a bit before upgrading. So then, will I be upgrading to Windows 11? I can’t do it even if I wanted to,  because the CPU in my main desktop PC is Zen 1 based, which isn’t currently officially supported  by Windows 11, and that’s because I’ve got the Threadripper 1950X.

For now, I’m waiting to see  what Zen 3 Threadripper brings to the table, and when I eventually upgrade my whole PC I’ll  probably go to Windows 11 at the same time. But that’s not likely to be until later this year  or early next year.

Even if I did have newer hardware that could run Windows 11 right now,  personally I’d hold off and wait. My impressions of Windows 11 are that it’s the same as Windows 10  in a lot of ways just with a fresh coat of paint, and it’s a combination of both minor bugs that  I’ve found while testing and those performance issues that would hold me back from upgrading. Those just aren’t the type of problems that I want to deal with on a production system that  I’m using every day.

Stability and to a lesser extent performance are far more important to me  personally than having the latest features, and I probably won’t be swapping all of my laptop  testing from Windows 10 to 11 until maybe early next year, but we’ll see. I just want  to be sure that the majority of issues are resolved first. I do always try to specify  the version of Windows in use in my testing, so just keep an eye out in the review videos.  Even when I do eventually make the transition over to Windows 11 for all of my benchmarking,  I’ll probably be doing it with core isolation disabled.

I don’t know though,  I need to have a bit more of a think about it, because with it enabled on the one hand you’re  getting worse performance than what we had available in Windows 10 but then on the other  hand if it’s going to be on by default in Windows 11 then doesn’t it make sense to test what most  people would have?

I’m not sure, but it’s a tough question that I need more time to think about.

So with all that in mind, let me know if you’ll be changing to Windows 11, staying on Windows 10, or  changing to Linux. Make sure you’re subscribed to the channel for future laptop content like this,  come and join me and the community in Discord and get behind the scenes videos  by supporting the channel on Patreon, otherwise for now you can check out one  of my other laptop videos over here next, I’ll see you in one of those..

Leave a Reply

Your email address will not be published. Required fields are marked *