Simple Windows 11 vs. Ubuntu 22.04.1

[Guest]

I noticed Ubuntu to be incredibly fast and based on recent extensive published benchmarks, I wanted to know how fast my processor was between Ubuntu vs. Windows.

I got a consistent 6% to 7% improvement in performance. Of course some may say I did a simple test and that Geekench is bla, bla, bla... But that's maybe related to CPU scheduling.

 

When, if ever, will Windows close the gap? What do you guys think this happens consistently?

Because more than 5% is really nice when pushing the hardware. Also my PC was cooler overall. 

 

Tests on Ryzen 5700G

[Good Bot, Bad Bot]

I am not seeing how a 5% benchmark difference really means much for real time results. Ubuntu? If using any Snap apps they will of course be slower to open than a Windows app.  I believe GPU performance is generally faster on Windows so there is that.

My main machine runs Linux and there is many reasons for that but not for any small benchmark advantage.

Edited December 8, 2022 by Good Bot, Bad Bot

[shockz]

You might notice a few extra minutes of battery life on Ubuntu, but that's probably an almost imperceptible difference... at least doing regular day to day tasks. 

[Guest]

On 08/12/2022 at 17:16, Good Bot, Bad Bot said:

I am not seeing how a 5% benchmark difference really means much for real time results. Ubuntu? If using any Snap apps they will of course be slower to open than a Windows app.  I believe GPU performance is generally faster on Windows so there is that.

My main machine runs Linux and there is many reasons for that but not for any small benchmark advantage.

Kernel 6.0 has a 15% advantage on the newer Ryzen 7000. We're not anymore on those days where 50% gains were the norm. For a software piece, that's a lot.

On 08/12/2022 at 17:25, shockz said:

You might notice a few extra minutes of battery life on Ubuntu, but that's probably an almost imperceptible difference... at least doing regular day to day tasks. 

Well, that means improvement in general for a free OS and not coming from one of the biggest companies in the world.

[Good Bot, Bad Bot]

On 08/12/2022 at 15:50, LuisMazza said:

Kernel 6.0 has a 15% advantage on the newer Ryzen 7000. We're not anymore on those days where 50% gains were the norm. For a software piece, that's a lot.

I suppose that is more significant for the few Linux users that would have a Ryzen 7000 machine. My point is even if Linux was 15% slower than Windows i would still be using it for my primary machine.

[Guest]

On 08/12/2022 at 17:16, Good Bot, Bad Bot said:

I am not seeing how a 5% benchmark difference really means much for real time results. Ubuntu? If using any Snap apps they will of course be slower to open than a Windows app.  I believe GPU performance is generally faster on Windows so there is that.

My main machine runs Linux and there is many reasons for that but not for any small benchmark advantage.

Also, 700 points on Geekbench is the single core performance of a Macbook Air from 2018, so... I also had a media player machine with 200 points on single core and 100 points is 50% of that performance. Just so we can see where we are.

[+Matthew S. Subscriber²]

Here's the thing, benchmarks are all fine and dandy, but they do not reflect real world performance.  The NT Kernel, XNU Kernel (macOS, iOS, etc...) and Linux Kernel are fundamentally different architectures, they are ALWAYS going to have performance gaps, advantages and shortcomings depending on what you are using the platform for.

[devHead]

On 08/12/2022 at 13:16, Good Bot, Bad Bot said:

I am not seeing how a 5% benchmark difference really means much for real time results. Ubuntu? If using any Snap apps they will of course be slower to open than a Windows app.  I believe GPU performance is generally faster on Windows so there is that.

My main machine runs Linux and there is many reasons for that but not for any small benchmark advantage.

Actually, I find opening Snap packages only slow on the first start.  After that, I notice no lag between, for example, the Firefox Snap and the Debian package.

[binaryzero]

OS wars, 90s is calling back...

[Mindovermaster Global Moderator]

Don't compare Windows to Linux, it's two different beasts. (I mean, the window or the penguin, you be the judge)

Each work in different ways in different scenarios.

As Matthew said, benchmarks are to be a guideline, not "this CPU is better because it runs 0.1Ghz higher than that CPU!" There are several ways in which they work. One benchmark doesn't tell us anything..

[Guest]

On 09/12/2022 at 03:20, Mindovermaster said:

Don't compare Windows to Linux, it's two different beasts. (I mean, the window or the penguin, you be the judge)

Each work in different ways in different scenarios.

As Matthew said, benchmarks are to be a guideline, not "this CPU is better because it runs 0.1Ghz higher than that CPU!" There are several ways in which they work. One benchmark doesn't tell us anything..

Yeah, so then we should do what Microsoft is doing: ignore other competitor's advantage because we are the standard.

Says Polaroid 👍

[adrynalyne]

On 09/12/2022 at 00:46, LuisMazza said:

Yeah, so then we should do what Microsoft is doing: ignore other competitor's advantage because we are the standard.

Says Polaroid 👍

You are playing benchmark games. Come back when you have perceivable real world comparisons to show. 
 

This is coming from someone that uses all of the three big OSes and likes Linux the most. 

[Guest]

On 09/12/2022 at 04:53, adrynalyne said:

You are playing benchmark games. Come back when you have perceivable real world comparisons to show. 
 

This is coming from someone that uses all of the three big OSes and likes Linux the most. 

Geekbench measures raw processor/memory performance.  It's simple and effective. Many other tests were run over the internet, and they all show that Linux has an advantage regarding "let the system reach its full potential". So, benchmark games are just a synthetic way of saying that regarding raw processor performance, Linux is better and even better with newer processors. Microsoft is just not optimizing its kernel as fast as Linux does.

[adrynalyne]

On 09/12/2022 at 01:07, LuisMazza said:

Geekbench measures raw processor/memory performance.  It's simple and effective. Many other tests were run over the internet, and they all show that Linux has an advantage regarding "let the system reach its full potential". So, benchmark games are just a synthetic way of saying that regarding raw processor performance, Linux is better and even better with newer processors. Microsoft is just not optimizing its kernel as fast as Linux does.

GeekBench is synthetic like you said. Synthetic. Not real world. I’ve had your performance improvement percentages differ from one test to the next. It’s a game. 
 

 

[Nick H. Supervisor]

On 08/12/2022 at 19:43, LuisMazza said:

I got a consistent 6% to 7% improvement in performance.

When, if ever, will Windows close the gap?

Because more than 5% is really nice when pushing the hardware. Also my PC was cooler overall.

As others have said, benchmark tests are indicators, but shouldn't be used for real-world scenarios. What kind of work are you doing where you need that 5% difference?

To answer the question though, Microsoft won't close the gap because they don't need to. When creating their OS they probably only use their previous OS as a reference point, I would be surprised if they compare it against another company's OS (at least, as far as performance goes). But I could be completely wrong about that.

[Guest]

On 09/12/2022 at 08:17, adrynalyne said:

GeekBench is synthetic like you said. Synthetic. Not real world. I’ve had your performance improvement percentages differ from one test to the next. It’s a game. 
 

 

Yes, but they variate 1% in the same OS. 7% variation is at least 3 times the variation of mean deviation of 2% variation. So that is statistically 3 times more significant.

[Guest]

On 09/12/2022 at 08:26, Nick H. said:

As others have said, benchmark tests are indicators, but shouldn't be used for real-world scenarios. What kind of work are you doing where you need that 5% difference?

To answer the question though, Microsoft won't close the gap because they don't need to. When creating their OS they probably only use their previous OS as a reference point, I would be surprised if they compare it against another company's OS (at least, as far as performance goes). But I could be completely wrong about that.

Now Windows is basically free for the end user who previously had a license, so what comes next? Steam Deck uses Linux, Steam is on Linux, if Microsoft loses the gaming community, which is highly focused on performance, Windows will become a niche OS for companies and lazy old people.

[adrynalyne]

On 09/12/2022 at 01:27, LuisMazza said:

Yes, but they variate 1% in the same OS. 7% variation is at least 3 times the variation of mean deviation of 2% variation. So that is statistically 3 times more significant.

It varies more than that. 

[Guest]

On 09/12/2022 at 08:32, adrynalyne said:

It varies more than that. 

Not more than 2% on my tests, unless I change BIOS settings. I also tested on Windows 10 and Windows 11 Dev.

Thus you don't care about those tests, how would you know?

[adrynalyne]

On 09/12/2022 at 01:36, LuisMazza said:

Not more than 2% on my tests, unless I change BIOS settings. I also tested on Windows 10 and Windows 11 Dev.

Thus you don't care about those tests, how would you know?

Not sure how or why you equated my lack of caring about benchmarks to not knowing about them…

 

I use them when making hardware or settings changes out of curiosity more than anything else.

 

A few percentage points is an imperceptible difference in the real world. 

[Nick H. Supervisor]

On 09/12/2022 at 08:30, LuisMazza said:

Now Windows is basically free for the end user who previously had a license, so what comes next? Steam Deck uses Linux, Steam is on Linux, if Microsoft loses the gaming community, which is highly focused on performance, Windows will become a niche OS for companies and lazy old people.

Well hang on. Using your argument of "Windows is basically free" why would people that have used Windows for all these years consider switching? For a neglible 5% performance boost? If you want to tell me that the 5% difference affects gamers, could you please provide an example from your own machine, playing a game on Windows and then playing the same game on Linux, and tell me with real statistics how great a difference it made (for example the difference in FPS, RAM consumption, etc.)?

Also, think about the computer manufacturers. I know that Dell offers Linux on select laptops and workstations, but the default is still Windows. And most people I know will just go with that rather than going through the steps to replace it with Linux.

I like Linux. I use Linux. But realistically this argument is as pointless today as it was in the 90's.

[Guest]

On 09/12/2022 at 08:55, Nick H. said:

Well hang on. Using your argument of "Windows is basically free" why would people that have used Windows for all these years consider switching? For a neglible 5% performance boost? If you want to tell me that the 5% difference affects gamers, could you please provide an example from your own machine, playing a game on Windows and then playing the same game on Linux, and tell me with real statistics how great a difference it made (for example the difference in FPS, RAM consumption, etc.)?

Also, think about the computer manufacturers. I know that Dell offers Linux on select laptops and workstations, but the default is still Windows. And most people I know will just go with that rather than going through the steps to replace it with Linux.

I like Linux. I use Linux. But realistically this argument is as pointless today as it was in the 90's.

7% is what I got for a 3 y/o processor. The newer processors have 15% better performance. 

Anyway, these Neowin forums discussion are basically a waste of time.  Everything is nah... it's always being like that... nah... no one cares. 

[Nick H. Supervisor]

On 09/12/2022 at 09:10, LuisMazza said:

7% is what I got for a 3 y/o processor. The newer processors have 15% better performance. 

Anyway, these Neowin forums discussion are basically a waste of time.  Everything is nah... it's always being like that... nah... no one cares. 

You're still using percentages. Can you provide me with information like I asked for above?

It's not that no one cares. I would care if I had real-world examples to look at rather than benchmark tests. But this puts the onus on you to support the idea that Microsoft needs to close that 5% gap. I've already explained why they don't need to.

[Guest]

On 09/12/2022 at 06:25, Nick H. said:

You're still using percentages. Can you provide me with information like I asked for above?

It's not that no one cares. I would care if I had real-world examples to look at rather than benchmark tests. But this puts the onus on you to support the idea that Microsoft needs to close that 5% gap. I've already explained why they don't need to.

https://www.softwaretestinghelp.com/linux-vs-windows/#Linux_and_Windows_Performance_Comparison

Linux And Windows Performance Comparison

The fact that a majority of the fastest supercomputers of the world that run on Linux can be attributed to its speed. Linux has a reputation for being fast and smooth while Windows 10 is known to become slow and slow over time.

"Linux runs faster than Windows 8.1 and Windows 10 along with a modern desktop environment and qualities of the operating system while Windows is slow on older hardware.

Talking about the core capabilities of OS like thread scheduling, memory management, i/o handling, file system management, and core tools, overall Linux is superior to Windows."

 

https://www.itpro.com/operating-systems/24841/windows-vs-linux-whats-the-best-operating-system/page/0/2

 

 

Windows vs Linux: Performance

Although Microsoft's flagship operating system is known for many things, it isn’t widely renowned for being lightweight and quick at the sight of a few other platforms. It can, in many machines, seem bloated and sluggish, and must be properly maintained or it’ll end up feeling too slow and out-of-date.

Linux, on the other hand, is known to be much quicker, largely due to the core of Linux being less demanding than that of Windows, with efficiency gains that some distros have provided, over time, to the performance levels. Some distros have taken these efficiencies a step further by binning a few of the flashier elements of the user interface (UI). This may create an unappealing, yet rapid experience for power users. It also means that Linux might be ideal for those with older machines that might not be able to handle the latest software.

Although you might be able to strip back Windows to the most basic elements to ensure your machine runs as smoothly as possible, Linux holds the advantage if efficiency is what you’re looking for.

[Nick H. Supervisor]

On 09/12/2022 at 11:02, LuisMazza said:

https://www.softwaretestinghelp.com/linux-vs-windows/#Linux_and_Windows_Performance_Comparison

Linux And Windows Performance Comparison

The fact that a majority of the fastest supercomputers of the world that run on Linux can be attributed to its speed. Linux has a reputation for being fast and smooth while Windows 10 is known to become slow and slow over time.

"Linux runs faster than Windows 8.1 and Windows 10 along with a modern desktop environment and qualities of the operating system while Windows is slow on older hardware.

Talking about the core capabilities of OS like thread scheduling, memory management, i/o handling, file system management, and core tools, overall Linux is superior to Windows."

And at no point have I disagreed with that statement. I agree that Linux is faster than Windows, not once have I said otherwise. But your question was, "when are Microsoft going to close that 5% performance gap?" and my response is that they won't because they don't need to, and then I asked you to provide real-world statistics (FPS, RAM consumption etc.) when playing a video game on Windows and then on Linux.

Of course a supercomputer will want to get every bit of performance out of the system, but I don't know anyone that has a supercomputer in their home for their day-to-day tasks...I know some guys over in CERN that have access to supercomputers, but they are hardly your average user.