Small Windows 7 vs Windows 10 face off

I`ve been dualbooting Win7 and Win10 for few days now and used this opportunity to directly and
objectively compare their performance.
Test conditions:
- Windows 7 Ultimate 64bit with Simplix UpdatePack v19.1.12 slipstreamed directly to ISO;
- Windows 10 Pro for Workstations 64bit with all WU updates installed (v17763.316);
- clean installations, default state of both systems, "out of the box" experience without any tweaks;
- everything hardware side was absolutely the same (P870KM1-G, 7700K, single 1080, 64GB 3000MHz DDR4 RAM);
- gpt partition table, UEFI system both;
- every BIOS option was absolutely the same;
- CPU undervolt (-120) and GPU overclock (+202 core) the same;
- every driver was exactly the same (even touchpad or bluetooth one lol), Nvidia was v417.71;
- no G-Sync, no Vsync, Nvidia max performance, Windows max performance power plan;
- for each test every graphical option was standardized (3DMark) or maxed out and identical between both Windows, all 1080p (except 720p free Catzilla);
- for games with DX12 support, DX11 was chosen for the obvious reason (but DX12 was benched too for science);
- all tests were ran 3 times directly one after another to include possible thermal throttling in results (Pascal clock oddity etc), numbers you see below are arithmetic average of those three runs;
- all tests were ran in offline mode with the same ambient temp.

Basically EVERYTHING was the same except Windows version.

Games used are from different eras and developers, built on many different engines by different teams to get wide and differential data. So how an oldie OS that debuted in 2009 holds up against its younger sibling from 2015? Results are interesting to say at least.



Numbers shows what we all know and experience since Windows 10 landed.
Cases where Windows 7 is better than 10 are numerous and that shouldn`t be the case with newer and "better" OS - you should expect to see Windows 10 outperform old Windows 7 in all or almost all tests.
To make some generalization and inference from synthetic benchmark results I would say that in tasks where CPU is more important Windows 7 performs better; tasks more GPU bond are better managed by Windows 10. At that time I was thinking about two nice and simple recommendations: You are mainly a gamer? Go for Windows 10! You are generally "everyday" PC user (Office, web, personal data, etc)? Go for (or stay on) Windows 7!

But this isn`t true anymore when we look at some games benchmark - in 17 tests Windows 7 was better 10 times (!) and Windows 10 only 5 times (quite shocking isn`t it?) with 2 draws (exactly the same scores). Windows 10 DX12 showed worse results than DX11 except one and only one case (Hitman), so generally it`s rubbish still, as we all know. You should expect newer games to perform better on newer Windows 10 (developing environment, SDKs, tools, game engines etc targeted to Windows version of its time) but it`s not true: some newest game perform better on Windows 7 (for example Shadow of the Tomb Raider).
Include in comparison how Windows 10 treats your privacy, messed updates and their quality, UWP, drivers, bloat, how more and more it takes control of hardware from user hands etc and decide yourself which one wins this little duel.

There are more benchmarks, more games with inbuilt ones (but I won`t be buying them) to test. If you have similar comparison data, please share it.

 

What build is that? 1809? Mine is 1803 far more stable then 1809 right now.
And you don't mention if SSD or HDD or NVMe drive. Windows 10 is made for SSD or NMVe drives so not knowing if your using this hardware can skew the results as well. I think Windows was made for SSD and NVMe and not running it in the environment it was made for can give one results one is looking for as well. I have Windows 10x64 Pro 1803 on NVMe drive. And as anything the testing/bench mark program have been caught in past test to skew in-favor of one hardware or software as well. So anyone using such programs to test I take with gain of salt.

 

It`s 1809 with all updates applied. I have three 970 EVOs (NVMe), so that`s what was used in tests (system and benchmarks/games were ran from 250GB).

 

This is very sad. Equally sad is the fact that your results are similar to my own testing almost 4 years ago. Nothing has changed, except that Windows 10 physics (CPU) performance is slightly worse now than when they first released it; probably because it has more bloat now. They haven't learned anything, and they're not getting better at what they do. They focus on stupid gimmicks that add no value and strive to make a PC as dumb as a smartphone. The Redmond Retards have had almost 4 years to make Windows 10 better than Windows 7 and they still have not been successful. They are an abject failure, and so is their garbage OS. Windows 10 is not, and shall never be, actually worth a damn. Windows 8.X was also equally a worthless pile of crap OS.

 

it would not be late for Microsoft to come out with Service Pack 2 (with recent harware support) for Windows 7 and ditch Windows 10.

 

They should, and as long as they don't mess it up too much it would be a smart move for them.

 

Instead we got this :

https://www.zdnet.com/article/how-m...d-on-windows-7-cost-you-heres-the-price-list/

 

There's un-official SP2 called Convenience update which allows to skip older updates pre 2016 and allows updating to newest ones way faster than before.

 

People will simply use Win 8.1 or LTSC version of Win 10 with 10 years of support and maintenance w/o buying the license again and again. Once LTSC long servicing is over, upgrade or clean install to next Long servicing branch for tension free usage in enterprises. For home users, I'd suggest to use WUB and NSudo to block updates.

 

Making 10 good enough for Candy Crush, then MS has achieved their goal for this millenium.

 

I don’t like neither the candy crush operating system nor 8.1. Will somehow the Simplix Pack contain the security updates till 2023?

 

I actually hate Windows 8.X more than I hate Windows 10, which is saying very little because any number multiplied by zero returns a product of zero.

For those that care about security updates, hopefully Simplix Packs will continue to be updated and have them. I don't use most of them and I will continue using Windows 7 until I can no longer obtain drivers needed to keep the hardware functional regardless of whether or not there are any security updates.

 

Has anyone seen a similar comparison between Pro and Enterprise LTSC?

 

Well, I tried that too but with some nuances.
Downloaded newest 1809 based Win10 LTSC x64bit and butchered everything that can be butchered from the ISO file (used ToolKit v9.1 for it - each and every component that ToolKit allows to remove was removed). The idea was to create THE slimest Win10 installation, as bloat free as possible and compare its performance with default standard Win7. Once again everything was (hardware, config, software, drivers etc) the same but with one change: RAM was set to 3000MHz via BIOS custom profile (previously for Win7 and Win10 Pro for Workstations 2400MHz was used). Lets give LTSC a small handicap

Tests I did were few selected benchmarks/games where Win10 Pro for Workstations showed worse performance than Win7, to see how much (if any) gain is achieved from extremely lite Win10 install and RAM overclock combined. List is much much smaller (no more time) but it shows consistent results.



As you can see, we do have gains everywhere, in all tests and all scores but... it`s still not enough to beat stock Win7 performance. Extremely slimmed down OS to almost non usable state (in normal everyday use) AND significant RAM overclock is still not enough to close that performance gap between new and old OS. I think that some drastic change(s) must be done at low (kernel) level and OS structure to improve Win10 performance.
Wanted to outperform much older OS? Forget it dude...

I`m basically done with this face off, all I wanted to know is here

 

+rep for testing and the thread.