Ubuntu 24.04 Yields a 20% Performance Advantage Over Windows 11 On Ryzen 7 Framework Laptop (phoronix.com) 63
Michael Larabel reports via Phoronix: With the Framework 16 laptop one of the performance pieces I've been meaning to carry out has been seeing out Linux performs against Microsoft Windows 11 for this AMD Ryzen 7 7840HS powered modular/upgradeable laptop. Recently getting around to it in my benchmarking queue, I also compared the performance of Ubuntu 23.10 to the near final Ubuntu 24.04 LTS on this laptop up against a fully-updated Microsoft Windows 11 installation. The Framework 16 review unit as a reminder was configured with the 8-core / 16-thread AMD Ryzen 7 7840HS Zen 4 SoC with Radeon RX 7700S graphics, a 512GB SN810 NVMe SSD, MediaTek MT7922 WiFi, and a 2560 x 1600 display.
In the few months of testing out the Framework 16 predominantly under Linux it's been working out very well. With also having a Windows 11 partition as shipped by Framework, after updating that install it made for an interesting comparison against the Ubuntu 23.10 and Ubuntu 24.04 performance. The same Framework 16 AMD laptop was used throughout all of the testing for looking at the out-of-the-box performance across Microsoft Windows 11, Ubuntu 23.10, and the near-final state of Ubuntu 24.04. [...]
Out of 101 benchmarks carried out on all three operating systems with the Framework 16 laptop, Ubuntu 24.04 was the fastest in 67% of those tests, the prior Ubuntu 23.10 led in 22% (typically with slim margins to 24.04), and then Microsoft Windows 11 was the front-runner just 10% of the time... If taking the geomean of all 101 benchmark results, Ubuntu 23.10 was 16% faster than Microsoft Windows 11 while Ubuntu 24.04 enhanced the Ubuntu Linux performance by 3% to yield a 20% advantage over Windows 11 on this AMD Ryzen 7 7840HS laptop. Ubuntu 24.04 is looking very good in the performance department and will see its stable release next week.
In the few months of testing out the Framework 16 predominantly under Linux it's been working out very well. With also having a Windows 11 partition as shipped by Framework, after updating that install it made for an interesting comparison against the Ubuntu 23.10 and Ubuntu 24.04 performance. The same Framework 16 AMD laptop was used throughout all of the testing for looking at the out-of-the-box performance across Microsoft Windows 11, Ubuntu 23.10, and the near-final state of Ubuntu 24.04. [...]
Out of 101 benchmarks carried out on all three operating systems with the Framework 16 laptop, Ubuntu 24.04 was the fastest in 67% of those tests, the prior Ubuntu 23.10 led in 22% (typically with slim margins to 24.04), and then Microsoft Windows 11 was the front-runner just 10% of the time... If taking the geomean of all 101 benchmark results, Ubuntu 23.10 was 16% faster than Microsoft Windows 11 while Ubuntu 24.04 enhanced the Ubuntu Linux performance by 3% to yield a 20% advantage over Windows 11 on this AMD Ryzen 7 7840HS laptop. Ubuntu 24.04 is looking very good in the performance department and will see its stable release next week.
Who is waiting to switch? (Score:3, Interesting)
Re:Who is waiting to switch? (Score:4, Interesting)
I've only tried the original Half-Life trilogy so far, but Steam on Linux seems to work well for at least some games.
And I know I'm a techie... But getting a Mint ISO and installing it was no more difficult for me than doing the same with Windows.
Some things still require being comfortable googling for answers and then opening a shell window to follow instructions, I did that just today to fix a Bluetooth audio delay that is apparently common. But if you're able to do more than just accept whatever came pre-installed, it's probably time to try Linux out.
Re: (Score:2)
Get out a pen and paper and help out, lazy human!
Re:Who is waiting to switch? (Score:5, Interesting)
Steam has Proton, which because of the Steam Deck, it has significantly improved gaming on Linux. About the only things it doesn't work on are games that require kernel anti-cheat.
And while Proton only works with Steam (though its changes are pushed back into WINE), you can still run other stores using it since you can add any app you want and run it under Proton.
Re: (Score:2)
Proton works amazingly well. Between it and WINE, Linux gaming is pretty good (as long as your hardware configuration is good and fully supported by properly functioning and configured drivers)
I have an all AMD setup. In theory, it is the simplest and easiest method for getting good results; however, the reality is that having an Nvidia card ends up working much more smoothly and reliably... and you get the bonus of having AI processing power.
If I have hyperthreading enabled and running a stock Linux Mint d
Re: (Score:3)
Re:Who is waiting to switch? (Score:4, Insightful)
I've got 126 games from Steam. Haven't had any problem with them running on Linux with AMD card. The big notables are The Sims 4, that remake FF7 that came to the PC, the Cities Skylines I/II, Civ 4/5/6, SimCity 3000/4, Octopath Traveler (just I, have to save up for II), the entire Disgaea series, I did that Among Us game for a bit there, the Deponia series, For the King (just I again $$ for II), Half-Life (all of them), Portal (both), that Paper's Please game, Myst/Riven/and the rest, that newer Tomb Raider and the two games that came after it, a couple of Total War games, XCOM 1/2/Chimera Squad, Yakuza 0, Stellaris, all of the Shadowrun games on Steam, and RimWorld to name a few.
That's not to mention all the ones I've gotten from GoG like the King's Quest series or ones I use something like Lutris to handle like WoW (I stopped playing some time but got plenty of years out of it) and Heroes of the Storm (I unfortunately still play it from time to time).
But just outside of the games. I've got discord, zoom, slack, dropbox, and spotify all working as well. Plus I have Google calendar via Thunderbird and do most of the things that are MS Office in either Google Docs or LibreOffice when I just need a PDF of the document. I have VSCode and it connects to PUB400 just fine, so I can literally code COBOL and RPG with VSCode on an AS/400. And eventually for the AS400 stuff, since IBM bought RedHat, they're putting a RedHat OpenShift container inside versions 7.5 and up of the OS and you can access that container via a web browser. It's not quite up to IBM RDi level, but it's rapidly getting there. And that's another thing, IBM RDi is based on Eclipse, it runs on Linux if I needed it to. We just use Windows because I don't know, I leave that shit to the admins that deal with that. But for all purposes, if I really needed to, I could move all my work over to Linux. There's not really anything stopping me, hell even the VPN client the company uses has a Linux program.
So yeah, the person who you were replying to has a very dated view of Linux. Windows 11 sucks, especially with stupid ass copilot they've just added. Especially with the dumb shit like "sign up for a OneDrive, blah blah blah" that shows up on even my work computer. I'm tired of Edge being shoved down my throat. It's not a bad browser, but I want to actively choose it, not be chosen for me. Microsoft's Office suite is starting to get a bit bloated and the Excel plus AI bullshit can go to hell. Also the fact that python in Excel is ran on MS computers and not locally can fuck right off.
Also, my nephew needed Windows 11 installed on their computer, fucking shit wouldn't let me install Windows without signing him up for a Microsoft account. THAT CAN GO STRAIGHT TO HELL. And then comes to find out you have to build a bootup that bypasses the requirement or you pop a command prompt to enter a command in a terminal to bypass it.
Yeah, Microsoft is just fucking running Windows into the damn ground. Fewer people want this shit they're cramming into it. Linux comes with none of, is easy to setup, and in a lot of things (NOT ALL OF THEM of course so some people Linux won't fit) will fit the bill.
I use PopOS for my daily and I've yet run into a problem with anything for my "at home" tasks. I've also helped at least six other folks who were Windows only folks convert and they've had next to no problems. In fact the only problem one of them had was getting their steam deck to play nicely with PopOS and there was a post on the forums on Steam that went step by step on how to get it going.
I mean a lot of people are just needing web at this point. But there's also a lot of people who need the industry standard tools that they've received on Desktop and yeah those people aren't changing any time soon. But with the way Windows 11 has been going, I'm hard pressed to recommend it to anyone unless they have some specific need. Windows 10 if they must have Windows, but Linux in most every other case. Windows 11 is
Re:Who is waiting to switch? (Score:4, Insightful)
Some things still require being comfortable googling for answers and then opening a shell window to follow instructions,
On windows too - it's quite common having to open a shell, or more often regedit, or run binaries from dubious websites to fix some arcane problem.
Re: Who is waiting to switch? (Score:2)
Re: (Score:1)
Also 8 times out of 10 those same people who are open to a non-Windows OS are just taking a trip to their Apple store.
Re: (Score:2)
And with Windows 11 now attempting to look like a MacOS installation the transition is more likely to happen until they discover that the Mac world is a walled software+hardware garden that's really a use and throw-away environment.
Re: (Score:2)
I think people running Mac are looking for simplicity. They are not going to switch to Linux unless they are techie and Mac gives them heck.
Re: (Score:1)
I think people running Mac are looking for simplicity.
Yes, they are deluded.
They are not going to switch to Linux unless they are techie and Mac gives them heck.
Even if their Mac shits all over them, they will praise it. Mac OS has been getting less and less reliable since about the same time they stopped calling it Mac OS X. The Mac users I know have become less and less enthralled with it as advertised functionality becomes more likely to malfunction. In particular they are distressed by recent degradations in the performance and reliability of Time Machine backups, which is arguably the thing Apple most needs to get right [apple.com]. (If all else fail
Re: (Score:2)
Mac OS has been getting less and less reliable since about the same time they stopped calling it Mac OS X.
bullshit. some apps have gotten buggier and resource-hungry because they're mashing the mobile and desktop environments together, and yes, that sucks, but the OS itself (and the other apps) are fine.
also, like, you (as in you, drinkypoo) have been bitching about Mac OS X for as long as there's been Mac OS X, so this new talking point really just feels hollow.
haven't heard about time machine fuckups, do
Re: (Score:2)
also, like, you (as in you, drinkypoo) have been bitching about Mac OS X for as long as there's been Mac OS X, so this new talking point really just feels hollow.
I'm consistent so you don't believe me? Okay, sport.
haven't heard about time machine fuckups, do you have a link to anything other than the obligatory "mAKe sUre tHE cAbLe iS PLuggeD IN" page every company has?
https://www.google.com/search?... [google.com]
Re: (Score:1)
I've recently started using Macbook Air again after a long time. My previous two Macbook pros had hardware failures. So not a good experience. My previous Dell XPS lasted almost 7 years, although I did have to replace the battery thrice, the keyboard and wifi card once each. I got the Macbook Air primarily for battery life, no fans and instant wakeup (although I think Windows 11 also has instant wakeup now?). This time I have an extended warranty which I think will be equal in expense to what I spent on the
Re: (Score:2)
That's kind of harsh. There is something to say for 'simplicity' and mac certainly brings that to the table but not all users are looking for simplicity. I have been using linux since 1998 but got into Mac around 2018 to see what a 'refined' unix environment would be like. There is a lot to like about Mac that's to be sure, it is quite polished.
It is off putting when people call it a walled in garden, though in some sense it is true, as I download and install packages from brew. That 'walled garden' allows
Re:Who is waiting to switch? (Score:5, Interesting)
Folks with Windows for games probably aren't going to bother with Wine.
I've been a Windows to Linux waffler since I put Slackware 2.0 on a 386DX25 with 8MB RAM and 120MB ATA hdd, using Kernel 1.1.47 (thus dating the start of my Linux saga) with A, N, D, and enough of the X set to run Netscape 2.0. And on that system I played (besides the epic classics like Nethack) Doom and Abuse. I ran Windows 7 for some time because it was a great place to run most games, even most of the vintage ones, and a tolerable place to run other things. I ran Linux occasionally in VMware Player or from USB stick for tasks that Windows couldn't or wouldn't do gracefully.
Now I run Devuan 5, and I am having a fairly excellent experience gaming with a combination of Lutris, PlayOnLinux, Steam, and Proton-GE. I only have a Pinnacle Ridge (1600AF) and a 4060 16GB, but I only game at 1080p. I got the version with more VRAM for LLM stuff, and so if/when I do get a 4k monitor, the card isn't worthless. I am frequently surprised by how many games I actually can run with this combination. With the exception of games with Windows kernel DRM, by far the vast majority of them can be made to work well.
If I were only gaming, I'd probably be on Windows 10. But Linux now is a very viable place to do a lot of gaming, and thanks to work put in to support the Steam Deck, a lot of games will now run very well indeed. Publishers of older games are also putting in a fair bit of work to make games function on Linux today. The new Fallout 4 patch coming out (I know that game is old AF, but it has an extremely active community) is Steam Deck Verified, but the game has run at least as well on Linux as on Windows for years now.
I do sometimes indeed still use normal Wine, but more commonly I use Proton-GE. Try it out, it's impressive.
Re: (Score:2)
Steam wants to install an old version of a game because it's a native Linux build rather than install a new version that's Windows-only yet fully compatible with Wine.
Re: (Score:3, Informative)
From the game's page in your Library, click on: Manage (the gear) -> Properties -> Compatibility -> check the box "Force the use of a specific Steam Play compatibility tool" then take your pick from the list. Be ready to re-download the entire game and possibly lose all your saves.
Re: (Score:2)
Skill issue.
Re: (Score:2)
Most phones run Linux, so that's who.
Re: (Score:3)
Folks with Windows for games probably aren't going to bother with Wine.
Of course not, proton is a far better option.
And desktops were eaten by phones
Literally everyone I know has both a phone and a desktop and/or laptop PC and uses both on the daily for non-work related things. No the desktop wasn't eaten by the phone. People are actually interested in not staring at a 2"x5" display.
Yeah (Score:2)
pretty much.
I guess it's nice to have numbers (Score:3)
But didn't everyone interested in avoiding the Microsoft tax already know this? Typically Linux is "start with the minimum, add what you need" while Microsoft is "throw in everything (except a few things you really want that we charge extra for) even if it makes the computer uselessly slow"?
There's a guy out there stripping out the unnecessary stuff from Windows, he calls it 'Tiny 11' (like 'Tiny 10' before it). I can't say I noticed a difference in capability but it ran nicely on old hardware I had where the MS original installation wouldn't.
Re: (Score:2)
Concrete numbers always builds confidence.
And you've made positive arguments yourself, but the title of the post doesn't draw attention to any of them and likely will stop many from reading what you have to say.
Re: (Score:2)
I'm at least trying to hold out for the Windows 11 LTSC.
Re: (Score:1)
No, people did not know this. Outside a few select niches of heavily advanced computer experience, the common assumption is that Windows is faster and more secure, otherwise they'd never have achieved market domination. Most people in fact would kill themselves if they had to admit any other explanation was possible.
Re: (Score:3)
I think the benchmark is justified. It was made by a website specializing in Linux benchmarks, on a Open Hardware laptop designed for Linux, and that natively comes with a Windows 11 double boot. Comparing the performance of the two boot options is useful to the readership of Phoronix, who already are Linux fans, and might be interested in purchasing this particular laptop.
Re: (Score:2)
More interesting than the benchmark is the conclusion that Framework's Windows support is poor. If you are thinking of buying one to run Windows, it might be a good idea to wait for them to improve the drivers and firmware.
In most tests there is no advantage for either Linux or Windows when compared on well supported hardware. Which is hardly surprising, since performance depends 99% on the hardware and by extension drivers, and 1% on the OS scheduler and file handling performance. Unless you are looking at
Re: (Score:2)
By reading the different benchmarks from Phoronix, there are constant patterns:
* that Linux is 6%-20% faster than Windows on stock gaming-level hardware for CPU tests,
* Windows is faster than Linux on GPU tests
* linux scheduler makes all the difference
* Intel ClearLinux (optimized by Intel for high performance computing) is always faster than Ubuntu by a good bit (and destroys Windows)
* on one occasion Windows was faster on the new Intel AlderLake, and Linux became faster again after the kernel upgraded
* Ub
Re: (Score:2)
I also think the benchmark is justified, and an interesting one. But I'm struggling to trust it.
The first image in the article: https://www.phoronix.com/bench... [phoronix.com] ... shows the processor running at 3.80GHz on Windows 11, and at 5.29GHz under the two Ubuntu versions.
If that's accurate, than no wonder Windows 11 lost in most of the CPU bound benchmarks! (Please note, I'm about as anti-MS as they come... I'm not trying to make excuses for Windows here; But this could make the whole benchmark moot)
The final conc
Re: (Score:2)
Looks like they are showing the base clock for Windows and the boost clock for Linux [techpowerup.com]. The processor model was listed on the page you linked, so you could have found this out with google.
The final conclusion, "Ubuntu ... a 20% Performance Advantage Over Windows", is quite the logical and linguistic leap. That seems to be based on the number of benchmarks run, counting each as a win/loss, and doing averages based on that. There are a TON of java based benchmarks, and a lot of that could be down the to java interpreter implementation. IE: it's benchmarking the java on linux implementation against java on windows just as much, if not more so, than ubuntu versus windows on this laptop.
So what you're saying is you think Java is optimized for Linux over Windows? I could believe that, but where are your benchmark tests? How did you leave those out?
Re: (Score:2)
So what you're saying is you think Java is optimized for Linux over Windows?
No, I'm saying the article didn't provide enough information to know what it is actually reporting. There are numerous java interpreters, and they all perform differently, sometimes vastly different. But they ran up a bunch of tests under Java, and most of those came out with Ubuntu being in the lead by a bit. Then, they used the total count of tests to derive the 20% advantage.
The clock speed thing... yeah, I know 3.8GHz is the base and 5.1GHz is the boost. Why did they list those differently under Windows
Re: (Score:2)
But they ran up a bunch of tests under Java,
Could it be you only read the first test page? There are 4 test pages filled with dozen of tests. Only the first page addresses OpenJDK.
Then, they used the total count of tests to derive the 20% advantage.
Of course they did as there is no other way to summarize a multidimensional result into a headline. The result could be suspicious if was surprising. But it's not surprising.Linux is faster than Windows all year long, and while a difference of 20% is on the high side, it's not a first time.
The clock speed thing...
Some of these benchmarks come with an explanatory sentence, here an example from anot
Re: (Score:2)
Then, they used the total count of tests to derive the 20% advantage.
Of course they did as there is no other way to summarize a multidimensional result into a headline. The result could be suspicious if was surprising. But it's not surprising.Linux is faster than Windows all year long, and while a difference of 20% is on the high side, it's not a first time.
Do you think the headline "Ubuntu 24.04 yields a 20% performance advantage over Windows 11 on Ryzen 7 Framework laptop"
What other metric do you propose to summarize the tests into a single number? Of course if you know your workload like you're doing mostly Blender rendering or video transcoding or 7zip compressing, you just check that part of the benchmark. The average does not really represent anyone's workload, its value is an indicator than can be followed along years, OS upgrades, hardware changes.
How about, "Ubuntu out performs Windows 11 on 90 of 101 tests on the same Ryzen 7 Framework laptop"? Or the headline the article itself uses, "Ubuntu 24.04 Boosts Performance, Outperforming Windows 11 On The AMD Ryzen Framework 16 Laptop" (Notably, that does NOT say it outperforms by 20%; The data doesn't support that statement)
Ubuntu 24.04 was the fastest in 67 of 101 tests.
Ubuntu 23.10 was the fastest in 22 of 101 tests.
Windows 11 was the fastest in 10 of 101 tests.
In no way would I interpret and present
Re: (Score:3)
Re: (Score:2)
Sometimes I wonder why this particular website bothers to cover issues related to that particular crowd. They're such an insignificant minority!
Re: (Score:2)
Most .deb-based systems can upgrade from one distro release to a newer one, especially if the hardware is new enough to have a TPM that supports Windows 11.
I have to opposite problem: my x86-64 laptop is from 2016, so I would have to use a workaround to put Windows 11 on it. It's a lot easier to update Ubuntu LTS every other year.
Re: (Score:3)
i had ubuntu on my laptop, at some point I had to upgrade the distribution, as apt-get stopped working on an older version - and upgrade failed.
Ubuntu software updates can pull from a very large number of distribution mirrors. During setup, it will usually pick the fastest mirror near you and set that in your configs.
Once a release goes past end of life, the software is pulled from all those mirrors. That's why you could no longer update - it was trying to find the updates on servers that no longer host the updates. Here's the doc on how to upgrade installs that are past end of life: https://help.ubuntu.com/commun... [ubuntu.com]
Basically, you just edit one tex
Windows rot? (Score:1)
Not sure if that is still true today as I gather Windows is fair more stable than it was when I suffered with it.
For this performance test I woul
OK... (Score:1)
So on one machine configuration, he got a big performance boost with one version of Ubuntu.
How is this particularly interesting, unless it's duplicated across many hardware configurations, and also isn't just a momentary issue with his Windows 11 or drivers that'll be fixed sooner or later?
I mean, I HATE 11, but this just seems like more Linux fanboism.
Re: (Score:2)
So on one machine configuration, he got a big performance boost with one version of Ubuntu.
I'm not sure that's true. I think the title is misleading. AFAICT, the new version of Ubuntu came in first on 20% more of the benchmarks than the others, but that doesn't mean it's 20% faster.
Re: (Score:2)
Gotcha.
100% irrelevant (Score:2)
Intuitive software that makes use faster will have a vastly bigger improvement overall.
With a better UI if you can save 20 minutes on entering data into a spreadsheet, who cares if it takes 1 minute longer to process it.
This is why we will supply the best computer for the user, if they are more proficient in Windows, OSX, Linux then
The "idiot" at the keyboard (Score:3)
speaks to the 'tude of the Linux community?
Go ahead. Mark me Troll. Make my day.
Battery life? (Score:2)
Most people buying laptops care much more about that than raw performance.
I run Arch on my desktop machine, and Windows on my laptop for that very reason.
16th edition (Score:2)
It makes me wonder. They say you only need to upgrade the insides, but then they keep releasing new notebooks.
Re: (Score:2)
Does 20% matter? (Score:1)
Is a 20% benchmark advantage noticeable to an average laptop user? We're talking laptops and not servers, so most tasks will involve GUIs and internet traffic and backend servers. Does a 20% advantage just for the laptop processing matter, especially considering that the laptop is often not the bottleneck?
Perhaps the bigger question is what percentage of users can be accommodated on either platform. Some programs only work on either Windows or Linux. If a program doesn't work on a particular platform, t
What the benchmarks do not tell you (Score:2)
Re: (Score:2)
Is this 20 percent boost in CPU and IO-bound tasks really worth it when you lose a lot more performance in other key areas?
Most intensive tasks are I/O bound, and most of the ones that aren't are CPU-bound, so yes.
Is Microsoft Mining in the Background? (Score:2)
Could this be? (Score:2)
Could it be? The year of Linux on the laptop?