Windows 11 leaked build (apparently) offers much higher performance than Windows 10.

[WolframaticAlpha]

Keyword is "apparently"

 

Summary

After some benchmarks, a user of the windows 11 leak has shown that the OS is quite a bit faster and more efficient than Windows 10. Windows 10 has been ragged on since day one for being a massive resource hog in comparision to its previous iterations. The source showed fall in boot times, higher single core and multicore performance and higher scores for the GPU in 3Dmark. The source ran these benchmarks on bare metal, without any of the OEM software. It seems most of the improvements are coming from better power management, slightly better optimization etc.

Quotes

Quote

First of all, the boot time is shorter at 13 seconds instead of 16 seconds. That’s an impressive change to start with, and it’s a good sign for overall Windows 11 performance. One test that he ran was Time Spy in 3DMark, which jumped from a score of 6,872 to 7,613. That’s both a bump in CPU and GPU, with the GPU score increasing from 6,927 to 7,426, and the CPU score increasing from 6,573 to 8,886. In his storage test on CrystalDiskMark, that did well as well, although he first noted that this one should be taken with a grain of salt due to the capacity being used. When he ran the test later, the results were the same. For Geekbench, the single-core performance increased from 1,138 on Windows 10 to 1,251 on Windows 11, and the multi-core score increased from 6,284 to 7,444, so there are some impressive improvements there too.

Quote

Windows 11 Pro Build 21996.1 booted 18.75% faster than Windows 10 Pro Build 19043 (13 seconds versus 16 seconds).

That tallies out to a 9.74% better 3DMark score and 2.05% better clock speed, albeit at the cost of a 7.08% hotter CPU.

That equals 9.04% better single-core performance and 15.59% better multi-core performance at a 2.05% faster CPU clock speed while running at a 4.13% cooler CPU temperature.

My thoughts

First of all, take this with a grain of salt. The results can be manipulated, and since there are few published tests, hence your mileage may vary.

These results may also have been manipulated. 

But if Windows 11 does improve performance over Win 10, it might just indicate that 11 is to 10, what 7 was to vista. 

LTT please do a video on this!

 

Sources

Windows 11 apparently offers big performance improvements over Windows 10 (xda-developers.com)

Windows 11 appears to outperform Windows 10 by a huge margin | Windows Central

Windows 11 is apparently faster than Windows 10 (windowslatest.com)

[Helpful Tech Witch]

The origonal artical i could find with the iso also said thet it was much more rescource efficient, and was installable with only 2gb of ram

[BadBoyHaloCat]

The install was also much faster.

[Murasaki]

What really needs to be compared between 10/11 is running processes/services/tasks and other calls. As for the boot time that probably is due to it being a new installation.

[Tieox]

This is the kind of articles I've been waiting for. 

 

 

[NotTheFirstDaniel]

I installed it on an HDD and it's actually not as unbearably slow as Windows 10 was. Not anywhere near as good as Windows 8/8.1, but close to Windows 7, which makes it actually usable! I mean, it's not all that useful nowadays, most computers ship with SSDs (and even cheap laptops nowadays use slow sata SSDs instead of using eMMC like the Win8 days), but it's a good step in the right direction IMO.

[Forbidden Wafer]

First things first: has anyone ever benchmarked windows before and after disabling Defender/ETW traces/search/compatibility appraiser/telemetry/insider service/etc? 

 

Windows 10 slowness in HDDs is due to Defender Real-time protection + Defender filesystem minifilter. Disabling it is a pain in the ass, but your computer feels brand new.

[Guest]

18 minutes ago, NotTheFirstDaniel said:

I installed it on an HDD and it's actually not as unbearably slow as Windows 10 was. Not anywhere near as good as Windows 8/8.1, but close to Windows 7, which makes it actually usable! I mean, it's not all that useful nowadays, most computers ship with SSDs (and even cheap laptops nowadays use slow sata SSDs instead of using eMMC like the Win8 days), but it's a good step in the right direction IMO.

Did you forget about $200 laptops that apparently many schools supply and they have 32 GB emmc?

[Mnky313]

weird, I didn't notice any difference on my system. Both were clean installs on the same SSD. Only difference in Windows 11 doesn't have chipset drivers (it says it's an unsupported version of windows)

[GoodBytes]

I would like to remind people that Insider builds follows A/B testing. So, you may have things that others don't.

[NotTheFirstDaniel]

11 minutes ago, GoodBytes said:

I would like to remind people that Insider builds follows A/B testing. So, you may have things that others don't.

This build looks so close to RTM (no build string on desktop/no "Microsoft Confidential" watermarks) that I highly doubt that they would implement A/B testing this late.

 

All the features we don't see, like the new Settings and Store app, are most likely in an experience pack that isn't published yet.

[Mnky313]

4 minutes ago, NotTheFirstDaniel said:

like the new Settings

wait, are they making another settings app....
_{oh, oh no.}

[StDragon]

1 hour ago, Mnky313 said:

weird, I didn't notice any difference on my system.

I knew this rumored assessment was BS; and your own tests confirm it.

 

See, what most people don't realize is that software optimizations are handled pretty well already with compilers. I think the Intel compiler is the best one out there and even helps AMD CPUs too. In fact, Intel's Linux distro named "Clear Linux" was reported to be the fastest by ArsTechnica

 

I don't know if MS uses any Intel compilers, and maybe they do, but unless MS changes this process, I don't expect much performance difference between Windows 10 and Windows 11.

 

About the only thing MS could do to free up CPU cycles would be to reduce the bloat and background telemetry. Though we're talking fractions of a percentage change here.

[LAwLz]

54 minutes ago, StDragon said:

I don't know if MS uses any Intel compilers, and maybe they do, but unless MS changes this process, I don't expect much performance difference between Windows 10 and Windows 11.

 

About the only thing MS could do to free up CPU cycles would be to reduce the bloat and background telemetry. Though we're talking fractions of a percentage change here.

Microsoft uses their own compiler called MSVC. But yeah, I think a lot of people (who doesn't know the first thing about programming) out way too much emphasis in "optimizing" code. If we're seeing almost 10% higher performance then it's not because of code optimizations done to the OS. It might be less bloat running in the background.

 

Windows is so complex that you can't just optimize and suddenly get like 10% higher performance. If we're lucky we might see optimizations that makes one API like 10% faster, but then that API might only be used in 0.1% of API calls in a program and the real world difference is a 0.01% performance increase. 

 

Software isn't magic, and the common operations in modern OSes are already extremely optimized.

[Forbidden Wafer]

1 minute ago, LAwLz said:

Windows is so complex that you can't just optimize and suddenly get like 10% higher performance. 

Kinda right and wrong. You can, but depends on the workload.

 

Windows have notoriously bad performance allocating/deallocating small blocks of memory frequently, opening/closing files and spawning threads. That's why people don't use it for HPC.

[GoodBytes]

48 minutes ago, LAwLz said:

Microsoft uses their own compiler called MSVC. But yeah, I think a lot of people (who doesn't know the first thing about programming) out way too much emphasis in "optimizing" code. If we're seeing almost 10% higher performance then it's not because of code optimizations done to the OS. It might be less bloat running in the background.

 

Windows is so complex that you can't just optimize and suddenly get like 10% higher performance. If we're lucky we might see optimizations that makes one API like 10% faster, but then that API might only be used in 0.1% of API calls in a program and the real world difference is a 0.01% performance increase. 

 

Software isn't magic, and the common operations in modern OSes are already extremely optimized.

Agreed. But typically, you have the company willing to make the business move for have team to start optimizing things that doesn't make much sense (example: Optimizing API xyz MIGHT lead to a 0.2% improvement). But this is the type of thing that leads to, when all combined, a more notable performance improvement.

 

I heard that Microsoft has been doing ongoing optimizations as well, well before this leak, for the next version of Windows. Sadly, I could never really test this. This is because I don't recall old benchmark scores from my aged old PC, and I can't play games on it... it's 12 years old. AAA game don't even run at playable framerate at minimum... Ok it does run at like 720p, minimum, on some game like Fortnight... but when my Switch can deliver better graphics... I am just using this. All I know, is that if I buy, AMD most entry level Ryzen CPU under the 3000 models, these CPUs are faster in benchmark than my system, lol! For context, it was in my plans to upgrade my system entirely (monitor and all) last year.. but you all know the market. So, waiting that long, and Cyberpunk 2077 wasn't that good to push me in buying one today... might as well just wait for next gen stuff.

 

Anyway, we will have to wait and see. I am hearing a lot of mix messages since a while. Stuff like general OS optimizations, to specifically SSD optimizations, and AMD CPU optimizations, or optimizations that is more visible on old/weak system... it's all over the place. So, I decided to not mention anything. I don't want to mention things that turns out to be false.

[LAwLz]

34 minutes ago, Forbidden Wafer said:

Kinda right and wrong. You can, but depends on the workload.

 

Windows have notoriously bad performance allocating/deallocating small blocks of memory frequently, opening/closing files and spawning threads. That's why people don't use it for HPC.

If we're going to be pedantic then yes, there are sometimes big performance uplifts from optimizations. But that's not the typical scenario like what we're talking about here.

If you are running some very specific workload that basically only uses a couple of very specific APIs in Windows, then yes it might make a big difference. A game on the other hand, or like 90% of other applications that uses a wide range of APIs and a wide range of operations? You won't see any big performance jumps for that.

[StDragon]

If anyone is up for the challenge (LTT maybe), you can still download (Rufus) all the older Windows 10 ISOs and benchmark with each one just to see what improvements have been made, when, and at what level (CPU, GPU, Storage, and RAM utilization)

 

I'd do it myself by I don't have a spare rig to part with.

 

 

[WolframaticAlpha]

11 hours ago, Murasaki said:

What really needs to be compared between 10/11 is running processes/services/tasks and other calls. As for the boot time that probably is due to it being a new installation.

Both were new installations

10 hours ago, Mnky313 said:

weird, I didn't notice any difference on my system. Both were clean installs on the same SSD. Only difference in Windows 11 doesn't have chipset drivers (it says it's an unsupported version of windows)

Bare metal or passthrough?

 

It might be exaggerated though. Maybe the dev build had certain enhancements to help compile times etc.

We do this at my work. Strip out useless stuff, then put in the other junk needed for like edge installs.

7 hours ago, LAwLz said:

Microsoft uses their own compiler called MSVC. But yeah, I think a lot of people (who doesn't know the first thing about programming) out way too much emphasis in "optimizing" code. If we're seeing almost 10% higher performance then it's not because of code optimizations done to the OS. It might be less bloat running in the background.

 

Windows is so complex that you can't just optimize and suddenly get like 10% higher performance. If we're lucky we might see optimizations that makes one API like 10% faster, but then that API might only be used in 0.1% of API calls in a program and the real world difference is a 0.01% performance increase. 

 

Software isn't magic, and the common operations in modern OSes are already extremely optimized.

Or maybe they did optimize it..... for benchmarks

8 hours ago, StDragon said:

I knew this rumored assessment was BS; and your own tests confirm it.

 

See, what most people don't realize is that software optimizations are handled pretty well already with compilers. I think the Intel compiler is the best one out there and even helps AMD CPUs too. In fact, Intel's Linux distro named "Clear Linux" was reported to be the fastest by ArsTechnica

 

I don't know if MS uses any Intel compilers, and maybe they do, but unless MS changes this process, I don't expect much performance difference between Windows 10 and Windows 11.

 

About the only thing MS could do to free up CPU cycles would be to reduce the bloat and background telemetry. Though we're talking fractions of a percentage change here.

Clear Linux user here. The 'Clear Linux is faster' shtick, while true, is not that useful. Sure, systemd boot was much faster. It was cleaner. But running real time stuff didn't show that many improvements. Major improvement I saw was in using clang linux(not in gcc) and in compiling packages. I tested this with a same version package and got a 10-15% improvement over ABS.

 

I use clear because of the way swupd updates the system.

[Mnky313]

4 minutes ago, WolframaticAlpha said:

Bare metal or passthrough?

 

It might be exaggerated though. Maybe the dev build had certain enhancements to help compile times etc.

Bare Metal, Laptop specs are in my signature.

I wouldn't be surprised if there is a slight increase in performance when the consumer (or at least insider) builds of Windows 11 arrive but I doubt it will be close to 10% faster.

[WolframaticAlpha]

7 minutes ago, Mnky313 said:

Bare Metal, Laptop specs are in my signature.

I wouldn't be surprised if there is a slight increase in performance when the consumer (or at least insider) builds of Windows 11 arrive but I doubt it will be close to 10% faster.

Was there OEM powertuning and crap involved in any of the installs?

 

 

I would be happy even if the perf remains the same. I honestly expected windows to become slower, because when have seen a perf improvement? 2009?

[Kisai]

55 minutes ago, Mnky313 said:

Bare Metal, Laptop specs are in my signature.

I wouldn't be surprised if there is a slight increase in performance when the consumer (or at least insider) builds of Windows 11 arrive but I doubt it will be close to 10% faster.

 

Here's what I think. If Microsoft makes it so that you can not install Windows 11 on anything without AVX, that means that the minimum requirements moved up, and the AVX instructions are being used in the OS by default. 

 

This to me doesn't suggest that much optimization can be done in the OS since AVX isn't an optimization for general data, it's a floating point optimization. So it would affect cryptographic things or video decoding. So perhaps if Bitlocker was enabled, it would be invoked.

 

[NotTheFirstDaniel]

11 hours ago, Mnky313 said:

wait, are they making another settings app....
_{oh, oh no.}

I should've been more clear. New = redesigned. Its the same WinUI settings app we all know and love, but with a new coat of paint. It's apparently supposed to look similar to the settings page on Microsoft Edge. Which is good. The current settings app was designed for a touch interface first, but seeing how Microsoft is backtracking from this perspective with Windows 11, they need a settings app that reflects it.

[Doobeedoo]

I mean overall I'd expect at least a slight improvement in everything. I'll wait for official release build to see. 

[WolframaticAlpha]

9 hours ago, Kisai said:

 

Here's what I think. If Microsoft makes it so that you can not install Windows 11 on anything without AVX, that means that the minimum requirements moved up, and the AVX instructions are being used in the OS by default. 

 

This to me doesn't suggest that much optimization can be done in the OS since AVX isn't an optimization for general data, it's a floating point optimization. So it would affect cryptographic things or video decoding. So perhaps if Bitlocker was enabled, it would be invoked.

 

I don't see why they would put AVX requirement on Home users though. Maybe they should add it as one of those windows features that you need to enable(like hyperv)

 

Also, microsoft needs a good name for hyperv. It can be miscontrued as hy-perv. Also windows 11 sould have a better UI for adding and removing modules and other power user stuff. I try doing anything non-normie, and an interface from 2007 pops up.