Windows 10 vs. 8.1 vs. 7 Performance: Microsoft's latest OS put to the test

[ozbat]

I am waiting for some DirectX 12 titles to come out. I actually installed Windows 10 and rolled back to Win 7 as simply to many programs tended to freeze up for no reason. I figure by the end of the year enough users will have pointed out flaws and lets face it MS has a real financial interest in fixing issue as the gambled big on Win 8 and lost; they need Win 10 to suceed. Win 10 has some great features but when in comes to Windows today I am not going to be an early adopter. As for performance, at this time I noticed no real difference between 10 and 7 running a core i7 with R9 290x's in crossfire. That will change when directX 12 titles are released towards the end of the year and from what I have read the R9 290x benefits a good deal more than the GTX 980 in some tests {look it up, plenty of articles}. By the way I despise the required password login and hope that changes. Thanks for a great article, very comprehensive.

 

[Steve]

You missed the most important thing in the comparison... Battery timing for each windows on same laptop???

This is a desktop comparison. That would certainly be missing if we did a laptop/mobile device review.

Doing benchmarks on SSD tech that pushes the limits of their associated bus timing, is a complete waste of time. It's the chipset drivers, the storage controller drivers that are important ... not the OS per se and not synthetics.

The obvious improvements built into Win 10 that are noticable are not addressed by any

of the items tested. Figure out how to benchmark these and you have Win ! :

Memory management
Indexing management
Screen page management
etc
etc

Even DISM processing is dramatically improved.

Except we did see some differences here and it is important to make sure that the hardware is running optimally. Furthermore there isn’t a way to test the things you mentioned so that isn’t particularly helpful either.

The performance measurement is not a true one.

There should be a way to reflect to the hassle and the headache that Microsoft enforced people to live through with this upgrade.

My Lenovo twist was working perfect with 8.1. I was totally convinced that it was the best ever Microsoft OS.

I read this article, and with the free upgrade, and the new touch features Microsoft promised, I went through the upgrade .....

The performance suddenly became BAD, and kept going from bad to worst.

I checked almost every technical article to resolve problems, half of them never worked with me, and still waiting for answers

My tablet used to restart in 5 seconds, not 2 minutes and 18 seconds; this is until yesterday, and after 4 days of suffering and agony.

Even when I rolled back, 8.1 is never working the same ....

Some apps don't work ... most importantly Windows Update ... it still insists that I should upgrade to Windows 10 for FREE ... as this word is the catch that will convince everyone to become Microsoft Rat Lab.

I reached to a point where I signed a petition for Bill Gates to fire all executives in Microsoft and get back to behind the steering wheel of this company, because those guys are ruining everything.

Just wanted to share this with you all. Hopes you never face this

Backup, Format, Re-install. Like I said I have installed Windows 10 on over half a dozen different devices now and it has been flawless each time. This is probably why others are struggling to help you.

LOL. NEVER! SSD has life time issues, performance degradation over time and I don't need a drill on how to replace/reinstall the system.

You have to be trolling right?

 

[MoeJoe]

Doing benchmarks on SSD tech that pushes the limits of their associated bus timing, is a complete waste of time. It's the chipset drivers, the storage controller drivers that are important ... not the OS per se and not synthetics.

The obvious improvements built into Win 10 that are noticable are not addressed by any

of the items tested. Figure out how to benchmark these and you have Win ! :

Memory management
Indexing management
Screen page management
etc
etc

Even DISM processing is dramatically improved.

"Except we did see some differences here and it is important to make sure that the hardware is running optimally. Furthermore there isn’t a way to test the things you mentioned so that isn’t particularly helpful either."



You have got to be kidding me.
All of those things are measurable.
It appears as though Techspot needs to hire some competent folks and add them to the mix.
This isn't rocket science.

Look, it's obvious to most why you guys - and so many other sites - resort to your standard
list of tests. That in-and-of-itself doesn't make it highly valuable in intent. Sure, having those things put to benchmark shed some perspective on the transition but does absolutely nothing to quantify the experience at the detailed level. And furthermore, the marginal differences you report are completely lost in the noise of the sampling basis and the granularity of the tests themselves.

 

[Steve]

"Except we did see some differences here and it is important to make sure that the hardware is running optimally. Furthermore there isn’t a way to test the things you mentioned so that isn’t particularly helpful either."



You have got to be kidding me.
All of those things are measurable.
It appears as though Techspot needs to hire some competent folks and add them to the mix.
This isn't rocket science.

Look, it's obvious to most why you guys - and so many other sites - resort to your standard
list of tests. That in-and-of-itself doesn't make it highly valuable in intent. Sure, having those things put to benchmark shed some perspective on the transition but does absolutely nothing to quantify the experience at the detailed level. And furthermore, the marginal differences you report are completely lost in the noise of the sampling basis and the granularity of the tests themselves.

Well please by all means give us your testing methodology for looking at Indexing management on each OS and we will include it.

 

[JetFixxxer]

You sure it's a windows problem?
For me this was the easiest and most problem free upgrade ever. For all the desktops and laptops I upgraded. No problems whatsoever with drivers needed. Even the old Llano dual graphics laptop worked out of the box. And this one had its problems with drivers in the last couple of years.
BSOD? What? Not something that is supposed to happen on a stable setup. Haven't had any in years.

Something in this version of Win10 is causing my machine to BSOD. Of course it could be the drivers that were released for Win10 that could be causing my BSOD. Either I will find out whats' causing it or I will just do a clean install. I rather find what's causing it. I like troubleshooting and keep the brain sharp

 

[Steve]

Something in this version of Win10 is causing my machine to BSOD. Of course it could be the drivers that were released for Win10 that could be causing my BSOD. Either I will find out whats' causing it or I will just do a clean install. I rather find what's causing it. I like troubleshooting and keep the brain sharp

Do you have the latest GPU drivers installed?

 

[albert89]

I don't think this reviewer has heard of DX12.
Someone please let him know and stop with the excuses of not testing it.
You could have included Vista & XP as well but I think your more scared of M$.

 

[Steve]

I don't think this reviewer has heard of DX12.
Someone please let him know and stop with the excuses of not testing it.
You could have included Vista & XP as well but I think your more scared of M$.

Umm ... okay can you please test some DX12 games and let us know the results?

Also what does Vista and XP have to do with this and why are we scared on Microsoft? We openly admitted that we thought Vista was rubbish and have done so many times in the past.

 

[Aliener]

In the real world out side of the States, Microsoft windows has become irrelevant and it will become more so in July 2016 when they start charging annual subscription fees for every user and they have to pay for updates/upgrades, that's why M$ is moving towards a SaaS (Software as a Service) system Over the last couple of years M$ have spent over 20 billion dollars They bought Skype, They bought Minecraft, they bought Nokia, Now it's coming to pay back time, for windows users, That's why Microsoft are forcing W10 onto Windows Xp W7 users, under the disguise of a Free upgrade, So be warned your about to start paying through the nose for everything

Er, I live in the UK, and I'd say you're completely wrong about Windows being irrelevant. Maybe amongst the few people you hang out with? The whole industry is moving to SaaS and or cloud-based services, not just Microsoft. Get over it. So they spent a pile of cash buying up services for their cloud push. Your point being? Windows 10 is free for the life of the device it's installed on.

The problem with Linux is the fragmentation, where no one can agree on a standard distro. Sure you might think choice and free is great but that ideology works poorly in the real world. I will only develop for platforms where I can make money. Developers need to monetise their applications to make a return. With Linux, no matter what, someone out there will do a free GNU/FOSS port and that's what's expected from its principles. It's been at least 15 years since Linux was due to take over and wipe out Windows on the desktop but it's still stuck on low single digit percentages. Why? Only OS X has made some gains against Windows in the last 4 -5 years, and that's really because more Mac hardware is being sold. You could say Linux has taken over on the mobile front with Android but that's another story in itself. Maybe mobile will be the only relevant platform in the near future? Whatever technical advantages Linux might have over Windows today is only relevant to a minority of people.

I run into people like you on a weekly basis that just simply dislike Microsoft for wanting to make money, and then spout old BSOD jokes for a laugh because they've been in the industry for 20 years and can't see past a terminal console. Sometimes they proudly show me their Ubuntu or Mint desktops with only a bash on the desktop, and then open up Linux.conf in vi. See? Yup, thanks, looks great.

 

[MoeJoe]

"Except we did see some differences here and it is important to make sure that the hardware is running optimally. Furthermore there isn’t a way to test the things you mentioned so that isn’t particularly helpful either."



You have got to be kidding me.
All of those things are measurable.
It appears as though Techspot needs to hire some competent folks and add them to the mix.
This isn't rocket science.

Look, it's obvious to most why you guys - and so many other sites - resort to your standard
list of tests. That in-and-of-itself doesn't make it highly valuable in intent. Sure, having those things put to benchmark shed some perspective on the transition but does absolutely nothing to quantify the experience at the detailed level. And furthermore, the marginal differences you report are completely lost in the noise of the sampling basis and the granularity of the tests themselves.

Well please by all means give us your testing methodology for looking at Indexing management on each OS and we will include it.

God Job picking one thing and avoiding the bigger context. Predictable.
And there's nothing worse than tech site whose staff troll their own threads.

Management needs to hire some industry insiders instead of going with mediocrity.

 

[deemon]

On that subject, 3DMark features a DX12 API Overhead test but this is only useful for comparing different hardware as there isn't a DX11 version and it can only be run using Windows 10.

What are you talking about no DX11 version? The 3DMark I have (from Steam) has universal "API Overhead test" which compares DX11 single-threaded, DX11 multi-threaded, Mantle and DX12... and my results on my ancient hardware were:
DX11 Multi-threaded draw calls per second 1 101 416
DX11 Single-threaded draw calls per second 1 133 340
Mantle - API not supported
DX12 draw calls per second 7 033 904

http://www.3dmark.com/3dm/8163659?

 

[deemon]

Also what does Vista and XP have to do with this and why are we scared on Microsoft? We openly admitted that we thought Vista was rubbish and have done so many times in the past.

I for one was using my Vista (on that same ancient hardware) until last month and it worked superbly. So I completely and honestly do disagree with you and think Vista was not rubbish. Have been using Vista and Win7 side by side (at least 5 days a week on both) for past 5 years and honestly only thing 7 has over Vista is SSD TRIM functionality and slideshow desktop wallpapers. Now when I swapped the Vista computer OS to 10, it feels a bit more fresh and "new" (since I skipped 8 and 8.1 entirely) but on performance side there is no big difference between them either - maybe some +5% performance, which goes largely unnoticed even.

 

[Steve]

God Job picking one thing and avoiding the bigger context. Predictable.
And there's nothing worse than tech site whose staff troll their own threads.

Management needs to hire some industry insiders instead of going with mediocrity.

Feel free to present a testing methodology for any one of the three things you brought up, I was just most interested in the one I picked. I am not sure what bigger context I avoided by picking something your brought up. Who’s trolling who? If your next post is dodging dribble expect it to be ignored or possibly even deleted.

What are you talking about no DX11 version? The 3DMark I have (from Steam) has universal "API Overhead test" which compares DX11 single-threaded, DX11 multi-threaded, Mantle and DX12... and my results on my ancient hardware were:
DX11 Multi-threaded draw calls per second 1 101 416
DX11 Single-threaded draw calls per second 1 133 340
Mantle - API not supported
DX12 draw calls per second 7 033 904

http://www.3dmark.com/3dm/8163659?

Okay my mistake. Still that isn’t particularly useful for comparing OS performance or even DX12 performance at this point.

I for one was using my Vista (on that same ancient hardware) until last month and it worked superbly. So I completely and honestly do disagree with you and think Vista was not rubbish. Have been using Vista and Win7 side by side (at least 5 days a week on both) for past 5 years and honestly only thing 7 has over Vista is SSD TRIM functionality and slideshow desktop wallpapers. Now when I swapped the Vista computer OS to 10, it feels a bit more fresh and "new" (since I skipped 8 and 8.1 entirely) but on performance side there is no big difference between them either - maybe some +5% performance, which goes largely unnoticed even.

Well again my mistake, I must be the only one that thinks Vista was rubbish when compared to Windows 7. Brining up a Windows 7 vs. Vista argument clearly wasn’t the point of that comment but hey why not.

 

[foxrox]

Nice article. I did some similar testing. Results are visible here: http://I.imgur.com/JTVkiBM.jpg

Or, here: http://forum.notebookreview.com/threads/windows-10.762434/page-181#post-10055473

And these links might be useful... all run on the same machine:
Fire Strike Comparison:
http://www.3dmark.com/compare/fs/5534923/fs/5542453/fs/5562514
Sky Diver Comparison:
http://www.3dmark.com/compare/sd/3228887/sd/3230919/sd/3236119
3DMark 11 Comparison:
http://www.3dmark.com/compare/3dm11/10107409/3dm11/10110810/3dm11/10116680
3DMark Vantage Comparison:
http://www.3dmark.com/compare/3dmv/5314125/3dmv/5314452/3dmv/5315108

 

[DAOWAce]

with no issues whatsoever on the performance front.

Uh, what?

There's clearly performance degradation in many areas on w10 compared to w7. Almost the entirety of the rest are within margin of error, while only one or two being slightly faster.

Either bad article writing or bias, which considering the w10 review posted a week or so ago, I'm leaning towards the latter.

 

[foxrox]

If you have a laptop with NVIDIA GPUs, be really careful with WIndows 10. Here are a couple of links to look at before diving into uncharted waters.

Windows 10 Problems and NVIDIA Drivers: https://forums.geforce.com/default/...ith-windows-10-and-the-latest-nvidia-drivers/

Windows 10 Causing Display Problems: https://forums.geforce.com/default/...g-samsung-and-lg-notebook-lcd-display-panels/

 

[foxrox]

Uh, what?

There's clearly performance degradation in many areas on w10 compared to w7. Almost the entirety of the rest are within margin of error, while only one or two being slightly faster.

Either bad article writing or bias, which considering the w10 review posted a week or so ago, I'm leaning towards the latter.

Yeah, there are clearly performance issues with Windows 10. Thought my links would be useful to give a bigger picture perspective to those that care more about performance and stability than being a Kool-Aid drinking guinea pig for Micro$oft. It was a Preview for about a year, and it's still not 100% ready for the noob masses. Making it free sounds nice, but it may come at a cost that some are not ready for and not able to effectively deal with.

 

[Steve]

with no issues whatsoever on the performance front.

Uh, what?

There's clearly performance degradation in many areas on w10 compared to w7. Almost the entirety of the rest are within margin of error, while only one or two being slightly faster.

Either bad article writing or bias, which considering the w10 review posted a week or so ago, I'm leaning towards the latter.

"Performance degradation" what are you talking about. As I said in the article almost all the results were within the margin of error. The only test that really stood out was PCMark 7 and it heavily favoured Windows 10. Yes there were times when Windows 10 was slower than 7 and or 8 but it wasn’t by a significant margin. You have to expect a few teething issues with a new OS so we didn’t pounce all over those results and try to make something out of them, if that is bias then so be it.

Still you come to the conclusion that this is “either bad article writing or bias”, I’m not even sure what bad article writing is, so maybe it’s that. It certainly isn’t bias, I don’t have a contact at Microsoft and never have. Moreover what is the alternative? Just another Microsoft product that they were charging you more money for :S

Naturally experiences are going to vary but as I said with over half a dozen of my own PCs now running the new OS I really like it. So far I have run into a few bugs which were highlighted in the conclusion.

Uh, what?

There's clearly performance degradation in many areas on w10 compared to w7. Almost the entirety of the rest are within margin of error, while only one or two being slightly faster.

Either bad article writing or bias, which considering the w10 review posted a week or so ago, I'm leaning towards the latter.

Yeah, there are clearly performance issues with Windows 10. Thought my links would be useful to give a bigger picture perspective to those that care more about performance and stability than being a Kool-Aid drinking guinea pig for Micro$oft. It was a Preview for about a year, and it's still not 100% ready for the noob masses. Making it free sounds nice, but it may come at a cost that some are not ready for and not able to effectively deal with.

I think you will find your sporadic performance is a result of system drivers rather than the operating system itself. There is no way Windows 10 is that much slower than either 7 or 8 for gaming. We have already moved GPU benchmarking to Windows 10 and with a battery of 20 games we haven't noticed any real changes. That said we are yet to test SLI performance but I would be shocked if it is that much slower.

 

[foxrox]

Naturally experiences are going to vary but as I said with over half a dozen of my own PCs now running the new OS I really like it. So far I have run into a few bugs which were highlighted in the conclusion.

I think you will find your sporadic performance is a result of system drivers rather than the operating system itself. There is no way Windows 10 is that much slower than either 7 or 8 for gaming. We have already moved GPU benchmarking to Windows 10 and with a battery of 20 games we haven't noticed any real changes. That said we are yet to test SLI performance but I would be shocked if it is that much slower.

The Windows 10 UI is clearly better than Windows 8/8.1 and I "like" that part of it better. With a tile-free Start menu (Classic Shell or StartIsBack) it's pretty tolerable in terms of aesthetics. Once Big Muscle has AeroGlass for Windows 10 ready for released it will actually be attractive to look at after sufficient cosmetic tweaks to tone down the over-abused white and light-colored pastel cartoon theme.

If you look at the information that I posted, the area where Micro$oft has been a consistent disappointment after Windows 7 is CPU performance. Windows 8/8.1 CPU performance was poor compared to Windows 7 and Windows 10 is worse than Windows 8/8.1 was. In terms of graphics performance, there is little difference between the three and until DX12 games become common there is no compelling reason to choose one version over another where graphics performance is concerned. As you said in the review, it is within a margin of error. In terms of SLI performance, the comprehensive testing I did shows how that looks.

Of course, my biggest concern at this point is what Micro$oft or NVIDIA has changed to actually cause harm to multiple high-end notebooks. (See link previously posted.) That needs to be fixed. Whatever capabilities the OS or GeForce drivers were given in order to have the capacity to affect that kind of change at a hardware level is inappropriate and highly disturbing. That sort if thing is a liberty one of them assumed that is way overstepping boundaries.

 

[RIMot]

All these stat's are ok but which 7 did you use 7 pro or the lower versions?

 

[foxrox]

All these stat's are ok but which 7 did you use 7 pro or the lower versions?

If you are asking the author, please ignore my reply.

If you are asking me, Windows 7 Ultimate x64. Windows 8.1 Pro x64 and Windows 10 Pro x64 were also used in my comparison benchmarks.

 

[DAOWAce]

"Still you come to the conclusion that this is “either bad article writing or bias”...

It's because it was written "no performance issues whatsoever".

From the benchmark results in the article I can see that there are issues outside of the margin of error ones. Issues are issues, however small anyone deems them to be. That is what my problem with the last page of the article is.

I'd suggest changing it to something along the lines of "no significant performance issues".

We openly admitted that we thought Vista was rubbish and have done so many times in the past.

As a Vista, 7 and 8 user, I find Vista to be the best OS of the bunch, while 8 would make me throw the computer out the window.

May want to reinstall Vista alongside your 7 installation to see that it's nowhere near the "failure" that so many people think it was on initial release, and in fact does many things better than 7. Maybe, just maybe, you'll change your opinion of it afterward and educate others about it too.

As for XP, it's pretty ancient by now, but man do I miss hardware audio. Even the stock Windows sounds sounded so much better than they do in later Windows versions. I really want to get my old system setup again with a CRT so I can play old classics properly.. *sigh*

 

[JetFixxxer]

Do you have the latest GPU drivers installed?

Sorry for the late response been on vacation.

Yes I have the latest GPU drivers.

It's MSI Afterburner that's causing the issue. AB is causing issue with Crossfire setups. I BSOD and a friend of mine his system just freezes. I'm running 2 R280x and he is running 2 6970.

If I select extend overclocking limits and Unofficial overclocking mode in AB settings it BSOD my computer. I'm using it just for fan control.

Right now I get flickering as my GPU 2 just keeps going from 6% to 0% use. I disabled ULPS both in the registery and in AB. With the GPU 2 my machine doesn't hibernate.

I usually overclock my GPU, but not by much and this takes care of the flicking.

 

[TheBigFatClown]

Well again my mistake, I must be the only one that thinks Vista was rubbish when compared to Windows 7. Brining up a Windows 7 vs. Vista argument clearly wasn’t the point of that comment but hey why not.

I certainly would never use the words "rubbish" and "Windows Vista" in the same sentence. After the second service pack it was a very solid and stable operating system. In retrospect, it's the second best version of Windows Microsoft ever released to date in my personal opinion. Every Windows release since Windows 7 has been less than what I find acceptable.

I have the creepy feeling Windows 10 is nothing more than an attempt to lead all the sheep to be slaughtered at a later date and time.

 

[Steve]

I certainly would never use the words "rubbish" and "Windows Vista" in the same sentence. After the second service pack it was a very solid and stable operating system. In retrospect, it's the second best version of Windows Microsoft ever released to date in my personal opinion. Every Windows release since Windows 7 has been less than what I find acceptable.

I have the creepy feeling Windows 10 is nothing more than an attempt to lead all the sheep to be slaughtered at a later date and time.

Well that is where you are I disagree. IMO Windows 7 was a massive step up from Vista and 8 was better than 7 (minus the start screen which I fixed using Classic Shell). So far for me Windows 10 is proving to be better than 8.1 and 7.

https://www.google.com.au/webhp?sourceid=chrome-instant&ion=1&espv=2&ie=UTF-8#q=vista+was+rubbish