Wednesday, August 19th 2026

CachyOS and AlmaLinux Beat out Windows 11 in Workstation Benchmarks

by
Cpt.Jank
 Discuss (87 Comments)
With the advent of the Steam Deck and the bump in Linux adoption that we've seen since then, a number of Linux distributions have appeared that make tweaks to the Linux kernel in the hopes of gaining performance. CachyOS is one such Linux distribution, and it appears that those kernel performance tweaks are paying off, according to some recent testing by Phoronix.

In a comparison between CachyOS, AlmaLinux 10.2, Ubuntu 26.04, Fedora Workstation 44, and Windows 11, CachyOS consistently performed at or near the top of the list, with CachyOS in performance mode claiming first place in the geometric mean of the results of a workstation benchmark. Windows 11 Pro was right at the bottom of the chart, coming in 14.9% slower than CachyOS in performance mode.
It's worth reading the full write-up at Phoronix for the in-depth results, especially if you're using benchmarks to determine which operating system to use with specific software, because there are enough cases in which Windows scores a decent victory that an application-specific approach is warranted.

The tests comprised of a mix of CPU and GPU benchmarks on a $46,000+ HP Z4 G6i workstation PC with a 48-core Intel Xeon 678X Granite Rapids WS CPU, 128 GB of DDR5-6400 memory, and an NVIDIA RTX Pro 6000 Blackwell workstation GPU. First up was 3D rendering and modelling, which tested Blender and IndigoBench 4.4 in both CPU and GPU rendering modes. Here, almost surprisingly, given NVIDIA's reputation in the Linux community, CachyOS and AlmaLinux were both faster than Windows 11 in both CUDA and CPU compute. The biggest differences in the render workloads came from CPU-only rendering, where Windows 11 was up to 12% slower than CachyOS. Ubuntu and Fedora were both measurably slower than CachyOS, but only by about 3-4%.
Tides turned with Chaos Group's V-Ray 6.0 rendering engine, where RTX testing resulted in 6% more rays on Windows, while CUDA rendering resulted in a 2% advantage for Windows. In CPU rendering, Linux was once again ahead, with CachyOS leading the pack at 18% faster than Windows but only marginally faster than the other Linux distributions tested. Neatbench video benchmarking, GravityMark 1.87, Unigine Superposition 1.0, and OctaneBench also saw CachyOS pulling ahead, although in some tests, gains were minimal, and in Octane, only CachyOS in Performance mode and AlmaLinux managed to pull ahead of Windows. In FluidX3D fluid dynamics simulations, performance was largely identical, but Windows 11 did manage to technically edge out a lead, even if only by fractions of a percent.
OSPRay Studio and LuxCoreRender produced peculiar results, with Windows matching or edging out Linux in a number of tests but losing badly to all of the Linux distributions in others. 7-Zip compression benchmarks once again saw Linux beating Windows by a decent margin, although it was basically a wash between the different Linux distributions.
Source: Phoronix

Related News

Add your own comment

87 Comments on CachyOS and AlmaLinux Beat out Windows 11 in Workstation Benchmarks

#26
lexluthermiester
GeoffreyAThis could have been tested with an LTSC installation, VBS and other things disabled, etc.
While that's true, using Home with proper tweaks and customizations should have been all that was required to get an accurate benchmark baseline. Even with VBS enabled, which is the Windows default config, that baseline is a perfect gauge for comparison. As tested, Pro was used, which is not the version that most users encounter, though not uncommon.

The comments about the Windows config and stating Windows is crap do nothing about the validation of the results. Default Windows config and default Linux config give the most clear perspective for what the general public can expect from a fresh install. That is the most objective, impartial and valuable info to be considered.
Posted on Reply
#27
GeoffreyA
AleXXX666SMART gamers "migrated" from HOME to PRO at least long time ago, even from XP era LMFAO. They don't know IF they will need all the functions of "pro/business" editions yet they somehow understand that all Home versions are stupidly castrated:):D
GrimlingI remember migrating from 98SE (ignoring ME) to Windows 2000 Pro.
I felt so "PRO" at that time...
Made us feel professional and high-tech back in the day :)
lexluthermiesterWhile that's true, using Home with proper tweaks and customizations should have been all that was required to get an accurate benchmark baseline. Even with VBS enabled, which is the Windows default config, that baseline is a perfect gauge for comparison. As tested, Pro was used, which is not the version that most user encounter, though not uncommon.

The comments about the Windows config and stating Windows is crap do nothing about the validation of the results. Default Windows config and default Linux config give the most clear perspective for what the general public can expect from a fresh install. That is the most objective, impartial and valuable info to be considered.
That's fair. One can only hope that Microsoft slims the default out-of-the-box installation.
Posted on Reply
#28
FR3D1
Ninja Cat
GeoffreyAWhich it usually is by default.
Even if it does, it has no significant impact on modern machines;

Windows 11 & 10: Benchmarks on VBS performance loss, including HVCI; VBS (including HVCI) in Windows 11: Latest benchmarks on gaming performance loss
ComputerBase (de)
lexluthermiesteras VBS is a part of the defender runtimes.
Or disable virtualization in the BIOS, because without it, HVCI (VBS) remains non-functional.
Memory integrity and VBS enablement
BIOS Virtualization must be enabled
Personally, a few "synthetic" benchmarks wouldn't convince me to switch. I only recommend Linux to people who just want to do some basic office work. It works well for that. Incidentally, I like to refer to various Linux distributions as "FrickelOS." It’s like those countless heavily modified custom ROMs that don't work quite as well as expected. I’ve mentioned this elsewhere before: if all those developers pooled their efforts, Linux could really become something great. How about we focus on Arch Linux? I like to compare it to LineageOS.

Minimalist, and therefore good.

Towards the end of the year, I’ll probably use my recently acquired "Lexar NM790 2TB" for a dual-boot setup. My 23H2 (Windows 11 IoT Enterprise) is reaching its end of support. I categorically rule out a Windows comeback. I selected "WinBoat" for "Office". I really don't care about gaming anymore, and for niche hardware like my dedicated sound card, there are if you will other, sometimes better solutions available. At a premium price, of course. But so be it.

Level up...

Regards,
Fred.
Posted on Reply
#29
GeoffreyA
FR3D1Or disable virtualization in the BIOS, because without it, HVCI (VBS) remains non-functional.
I keep virtualisation off in the BIOS, and Hyper-V off in Windows. Without the latter, one can turn on virtualisation when needed without VBS kicking in.
FR3D1I categorically rule out a Windows comeback.
A year ago, yes. This year, there have been steady improvements and we have yet to see the more important ones, such as the rewriting of shell components in WinUI.
Posted on Reply
#30
lexluthermiester
FR3D1Or disable virtualization in the BIOS, because without it, HVCI (VBS) remains non-functional.
Memory integrity and VBS enablement
GeoffreyAI keep virtualisation off in the BIOS, and Hyper-V off in Windows. Without the latter, one can turn on virtualisation when needed without VBS kicking in.
A lot of power-users, myself included, do that.
Posted on Reply
#31
FR3D1
Ninja Cat
GeoffreyAsuch as the rewriting of shell components in WinUI.
I would prefer "under-the-hood improvements." In my opinion, these superficial changes are just there to reassure less tech-savvy users.

Regards,
Fred.
Posted on Reply
#32
GeoffreyA
FR3D1I would prefer "under-the-hood improvements." In my opinion, these superficial changes are just there to reassure less tech-savvy users.

Regards,
Fred.
I would too. But surely, those kernel and executive improvements are going in; they're just not visible to us, mostly.
Posted on Reply
#33
AleXXX666
GrimlingI remember migrating from 98SE (ignoring ME) to Windows 2000 Pro.
I felt so "PRO" at that time...
My first home PC was with ME, and you know, I don’t remember any major “crashes” every second, later when XP released we put it there also, and both worked OK for basic “preinstalled” games or simple casual games of that era, I think it was even offline at that time and we were watching movies on compact discs lol. Such a warm nostalgia:) :love: :D
Posted on Reply
#34
TheinsanegamerN
lexluthermiesterLook in a mirror before making comments like that. Linux IS getting very excellent, however, Windows 11 properly configured is also excellent.

CachyOS, Alma, Bazzite and many others are becoming very worthwhile alternatives to Windows. Better? That is highly subjective and depends on what you want to do with your PC.
If you have to butcher and carve up your OS to make it perform properly, then your OS is crap.
Posted on Reply
#35
JustBenching
TheinsanegamerNIf you have to butcher and carve up your OS to make it perform properly, then your OS is crap.
Turning VBS off isn't butchering anything. Someone on the other side of the extremist table could argue that linux not having it on by default makes it an unsafe swiss cheese OS.
Posted on Reply
#36
TheinsanegamerN
JustBenchingTurning VBS off isn't butchering anything. Someone on the other side of the extremist table could argue that linux not having it on by default makes it an unsafe swiss cheese OS.
Keep in mind that Lex's " properly configured" windows system is open that has had most of Microsofts apps like edge and defender stripped out. That is not a matter of just clicking buttons.
Posted on Reply
#37
GeoffreyA
TheinsanegamerNIf you have to butcher and carve up your OS to make it perform properly, then your OS is crap.
To be fair, LTSC is streamlined out of the box, and is not very different in spirit from, say, an optimised or minimalistic Linux distribution. Then, disabling VBS takes one BIOS setting, or turning off Hyper-V in Programs and Features (or whatever it's called today).
Posted on Reply
#38
JustBenching
TheinsanegamerNKeep in mind that Lex's " properly configured" windows system is open that has had most of Microsofts apps like edge and defender stripped out. That is not a matter of just clicking buttons.
I get it but i think the biggest issue is VBS itself, which - although in some cases has 0% performance drop - in some other cases it can be 10 or 15%. I don't think you should be comparing a custom tweaked OS but VBS ON / OFF should be tested for a proper comparison.

It's a very dishonest / misleading test to be done this way. People might get the wrong idea and go ahead and "upgrade" to linux. I've fallen pray to that a couple of years ago when I moved to bazzite for my laptop cause "it's lightweight, takes no resources, windows is bloated crap" and all that and it ended up killing baterry life almost twice as fast as windows did.
Posted on Reply
#39
GeoffreyA
Not sure how reliable AncientGameplays is, but Windows is holding up in gaming too:
Posted on Reply
#40
phanbuey
JustBenchingVBS add security though. It's not like it's a feature that does nothing. Then the article should have been Linux is slightly faster but a Swiss cheese in security.
Linux is by no means swiss cheese - there are a ton of privilege boundaries that are lightweight and effective and don't require wrapping your entire OS in a hypervisor layer on top of memory entropy and constant above ring 0 checks.

MS uses VBS as a brute force approach that has a massive overhead because they have to support an insane amount of legacy code (and let that legacy code do sketchy crap to preserve compatibility). The only way they can guarantee security is assuming the kernel is already compromised - that's not some super great genius security feature - it's a brute force throwing up your hands and going "F$@^ it, treat the kernel as already compromised and don't trust it, wrap it in a vm - performance be damned".
Posted on Reply
#41
lexluthermiester
TheinsanegamerNIf you have to butcher and carve up your OS to make it perform properly, then your OS is crap.
Your opinions are often the same and they mean about as much to me as an anthill in Africa.
TheinsanegamerNKeep in mind that Lex's " properly configured" windows system is open that has had most of Microsofts apps like edge and defender stripped out. That is not a matter of just clicking buttons.
Yet I didn't mention ANY of that in this thread. Context is important, something you frequently fail at. Are you ever going stop with your silly personal attacks?

Also, just an FYI, Linux is, by it's very design, high customized and customizable. It's the complete nature of the beast, so complaining about my methodology as applied to Windows and in comparison to Linux makes as much sense as a screen-door on a submarine.
Posted on Reply
#42
FR3D1
Ninja Cat
TheinsanegamerNthen your OS is crap.
To what extent does this not apply to countless Linux distributions?
TheinsanegamerNThat is not a matter of just clicking buttons.
That doesn't apply 100% to Edge unless @lexluthermiester lives outside the EU. Regarding Windows security: perhaps @lexluthermiester is running Windows 11 in a VM. In that case, that specific security feature wouldn't be installed without him even having to do anything.

By the way,
When I recently mentioned somewhere deep within TPU that Windows 11 might perform better in a VM, I didn't just throw that out there at random.

Regards,
Fred.
Posted on Reply
#43
JustBenching
phanbueyLinux is by no means swiss cheese - there are a ton of privilege boundaries that are lightweight and effective and don't require wrapping your entire OS in a hypervisor layer on top of memory entropy and constant above ring 0 checks.

MS uses VBS as a brute force approach that has a massive overhead because they have to support an insane amount of legacy code (and let that legacy code do sketchy crap to preserve compatibility). The only way they can guarantee security is assuming the kernel is already compromised - that's not some super great genius security feature - it's a brute force throwing up your hands and going "F$@^ it, treat the kernel as already compromised and don't trust it, wrap it in a vm - performance be damned".
Im not suggesting it is a swiss cheese. It was just an example to demonstrate the point. Linux does have it's own version of VBS, it's just that it isn't on by default.
Posted on Reply
#44
lexluthermiester
FR3D1That doesn't apply 100% to Edge unless @lexluthermiester lives outside the EU. Regarding Windows security: perhaps @lexluthermiester is running Windows 11 in a VM. In that case, that specific security feature wouldn't be installed without him even having to do anything.
Nope, I'm in the USA. I use a customized setup for my personal install of 11.

But I also highly customize my Linux installs, removing things I don't want installed and general debloating. Yes, Linux has bloat too! Surprise, surprise!
Posted on Reply
#45
CajunMoses
There's one thing about these comparisons that really bothers me. These tests are almost certainly on fresh installs. CachyOS is a rolling release. So, its performance may degrade and then plateau over time, depending upon how it's used. The performance of an immutable, atomic distro like AlmaLinux may actually improve over time. A point release like Fedora Workstation, Windows 11, or Ubuntu may gradually degrade until restored by its point release upgrade. It would be nice to these comparisons for systems that had been used as "daily-driver" systems for at least six months. Smart money would probably be on the atomic OS.
Posted on Reply
#46
lexluthermiester
CajunMosesThere's one thing about these comparisons that really bothers me. These tests are almost certainly on fresh installs. CachyOS is a rolling release. So, its performance may degrade and then plateau over time, depending upon how it's used. The performance of an immutable, atomic distro like AlmaLinux may actually improve over time. A point release like Fedora Workstation, Windows 11, or Ubuntu may gradually degrade until restored by its point release upgrade. It would be nice to these comparisons for systems that had been used as "daily-driver" systems for at least six months. Smart money would probably be on the atomic OS.
Windows has been a "rolling release" since Windows 8, so comparisons might shift, but also remain valid.
Posted on Reply
#47
_roman_
CajunMosesThere's one thing about these comparisons that really bothers me. These tests are almost certainly on fresh installs. CachyOS is a rolling release. So, its performance may degrade and then plateau over time, depending upon how it's used. The performance of an immutable, atomic distro like AlmaLinux may actually improve over time. A point release like Fedora Workstation, Windows 11, or Ubuntu may gradually degrade until restored by its point release upgrade. It would be nice to these comparisons for systems that had been used as "daily-driver" systems for at least six months. Smart money would probably be on the atomic OS.
The file systems are far more advanced than ntfs.

ntfs is worse as ext4. BTRFS wins over etx4.

Whataboutism, If there would be performance degradation, i would see it on my 21 year old gnu gentoo linux installation. I do not use systemd, but openrc.

I moved several times the full system from hardware to hardware. I would have seen an impact from
corsair mp 600 pro 1tb -> crucial p5 plus -> KC 3000 -> mp 600 pro -> kc3000

the impact I may argue what you hint out, are file system regression which do not exists on modern high advanced file systems like btrfs, waht windows has massive issues wiht lint files.
e.g. try to run ntfs on a hdd with several defective sectors or on a sata ssd. vs btrfs + zstd compression

kernel issues, but you rebuild the kernel, that is one of teh components which get swapped out several times a month.

most userspace gets switched out. old packages gets removed. there are hardly any packages which are older than 2 years on my box.

I saw performance impact with low quality SATA SSDs on my previous harware, a gaming notebook, especially with ADATA SATA SSDS and sandisk SATA SSDS compared with quality drives from CRUCIAL or transcend.

I would also see it when changing file system, or for the compile speed.

I had to repair NTFS this year or last year for my windows 11 pro. It was defective. That is hard fact. And no, it is not the kc3000 m2 nvme drive fault

Windows users are stuck with ntfs and do not know the benefits of advanced file systems or the beneifts of encryption or 15 yera old features like lvm2.
Posted on Reply
#48
Dukenukemx
Infy101The Phoronix comparison is abit flawed. According to their test setup, the Windows installation is not a clean one unlike the Linux installations:
  • It comes with OEM pre-installed bloat
  • It has all optional Virtualization Based Security features enabled, none of which exist on Linux
  • It also has optional Spectre/Metldown mitigations enabled while the Linux installations only have the default mitigations
On Windows VBS and Spectre/Metldown mitigations can really hurt performance.

Phoronix should have performed a clean installation of Windows so they could have compared the out-of-box performance of the operating systems.
Linus Tech Tips did a test and the bloat doesn't effect benchmarks very much. I also doubt that Phoronix left all the bloat running while doing their tests. Even still, wouldn't have made a difference if Phoronix forgot to remove the bloat and disable anti-virus.
_roman_I have a full list of CVE antimeasures active in my kernel. and my box is not optimised.

Shall i remove that and cross test? I could rip out all security measures, downgrade the motherboard uefi so it is faster etc.

Hardware gets slower because of the CVE fixes in UEFI and software. Taht also includes the kernel of teh operating system.
Keep in mind this is for a workstation which means you'll likely keep VBS on. I use Linux with mitigations=off because ain't nobody going to use CPU vulnerabilities for anything yet. You can do the same for Windows but realistically you will keep all your security measures enabled.
Posted on Reply
#49
MTS-C
TheinsanegamerNThat is not a matter of just clicking buttons.
One click button for me personally. I don't remove Edge though.
Posted on Reply
#50
AleXXX666
_roman_ADATA SATA SSDS and sandisk SATA SSDS compared with quality drives from CRUCIAL or transcend.
sandisk is WD.:rolleyes: Yeah quality of these cheapos are "russian roulette", but it's completely SAME for example for Kingston cheap SSDs, for example "the great" A400. But "transcend" associates for me only with cheapest garbage, at least their mem cards and USBs are so lol Crucial BX series isn't "great performer" also, and health degrading speed of MX series tells that their "quality" is very questionable. The only real "quality" remains.. you know... "pay for SATA ssd as for NVME"... Samsung:D
_roman_I had to repair NTFS this year or last year for my windows 11 pro. It was defective
for how many years of installation? And is it gone "defective" by user input (software installs-uninstalls) or "by itself"?:rolleyes: Can't remember I need to use FOR REAL "sfc scannow" besides BS that this is a "panacea" for every windows error lol
_roman_Windows users are stuck with ntfs and do not know the benefits of advanced file systems or the beneifts of encryption
there is ReFS you know.. But I haven't tried it as it can't be used for boot drive, and for other drives I'm happy with ExFAT as I can connect drive to any other OS machine without any extra "mess".:):D

I have used BitLocker encryption on Windows and if configured properly and with good hw it's pretty okay..:rolleyes::) But do I really need it? No. Better safe than sorry, so means better don't give your data even encrypted rather than give it for someone lol
Posted on Reply
Add your own comment
Nov 13th, 2026 22:48 CST change timezone

Latest GPU Drivers

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts