Wednesday, August 19th 2026

CachyOS and AlmaLinux Beat out Windows 11 in Workstation Benchmarks

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
Add your own comment

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

#1
Dukenukemx
This isn't news to anyone on Linux. The only difference is that his is done on a $46k workstation. You know, something people have in their basements.
Posted on Reply
#2
Exilarch
I don't think these benchmarks are fair. Windows 11 is at a serious disadvantage. Because Windows 11 is retarded.
Posted on Reply
#3
holyprof
I'm sure that M$ could strip all the unnecessary bloat/ads/telemetry/AI slop from Windows, making it a little bit faster. Probably not as fast as Linux, but at least not so embarassingly slow.
Well, one can dream.
By the way, I finaly installed Windows 11, due to new motherboard. Using just the Rufus options to remove CoSlop and telemetry, it's so much faster than the "stock" Windows 11 experience, almost as fast as Windows 10, but with better support.
Posted on Reply
#4
ISSEnterprise
If my memory is correct, best performance isn't Windows' advantage since its early days. The UNIX workstations it replaced still perform better in productivity. But Linux still have very long way to go before becoming a mature desktop operating system for non-programmers.
Posted on Reply
#5
Bigshrimp
Windows is a closed source slopfest of a marketing platform as an OS and Linux is an open source OS without the bloat and better performance. I remember Windows XP, when Microsoft actually developed a decent OS, then they screwed it all up because the line must go up at all costs. So they added a bunch of slop marketing and telemetry into their OS and was like this will win over our customers like any delusional mega-corporation would think.
Posted on Reply
#6
lexluthermiester
ExilarchI don't think these benchmarks are fair. Windows 11 is at a serious disadvantage. Because Windows 11 is retarded.
Look 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.
Posted on Reply
#7
Let3r
holyprofI'm sure that M$ could strip all the unnecessary bloat/ads/telemetry/AI slop from Windows, making it a little bit faster. Probably not as fast as Linux, but at least not so embarassingly slow.
Well, one can dream.
By the way, I finaly installed Windows 11, due to new motherboard. Using just the Rufus options to remove CoSlop and telemetry, it's so much faster than the "stock" Windows 11 experience, almost as fast as Windows 10, but with better support.
Yeah. I would probably use Windows over Linux if Microslop removed all the bloat, tracking, and AI slop. And maybe made it a as completely customizable as Linux.
Posted on Reply
#8
GeoffreyA
There's a big question mark on this test. Is VBS enabled? Which it usually is by default.
Posted on Reply
#9
Infy101
The 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.
Posted on Reply
#10
JustBenching
GeoffreyAThere's a big question mark on this test. Is VBS enabled? Which it usually is by default.
Exactly what I wanted to ask. But of course everyone jumped to "win 11 is crap". Ignorance is bliss.
Posted on Reply
#11
lexluthermiester
GeoffreyAThere's a big question mark on this test. Is VBS enabled? Which it usually is by default.
This is not an issue for anyone who has disabled Windows Defender as VBS is a part of the defender runtimes.
Posted on Reply
#12
Hecate91
A majority of Linux being faster by default vs. Windows is the amount of slop and bloat Windows comes with, VBS being on is a Windows problem and it being on by default is fair enough IMO.
Windows 11 is crap once you've tried Linux you see just how crap it really is,Linux being much more smooth to use without bloat, forced updates breaking things, ads, and spyware, although of course ignorance is bliss.
Posted on Reply
#13
LittleBro
CachyOS is an excellent Linux distro. It destroys SteamOS on Steam Deck.
Bazzite is fast, too, but there are restrictions, it's like closed Linux. Also, Bazzite is Fedora-based, while CachyOS is fork of Arch.
Posted on Reply
#14
qlum
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.
generally the case phoronix not a good benchmarker, often a bigger question is if the benchmarks he runs tell you something useful. Game benchmarks are especially questionable when he does them.

Do want to add the goal was to test catchyos so there was never a real focus on the windows benchmark, and flaws aside, generally in these tests linux outperforming windows is the expected outcome.
Posted on Reply
#15
lexluthermiester
Hecate91Windows 11 is crap once you've tried Linux you see just how crap it really is,Linux being much more smooth to use without bloat, forced updates breaking things, ads, and spyware, although of course ignorance is bliss.
You making this statement and ending it in that way is telling. It says you don't know how to setup Windows properly & securely. It's not exactly secret information.
qlumgenerally the case phoronix not a good benchmarker, often a bigger question is if the benchmarks he runs tell you something useful. Game benchmarks are especially questionable when he does them.

Do want to add the goal was to test catchyos so there was never a real focus on the windows benchmark, and flaws aside, generally in these tests linux outperforming windows is the expected outcome.
While this is partly true, the benchmarks do show that Linux is becoming very competitive with Windows, which is a VERY good thing!
Posted on Reply
#16
GeoffreyA
Hecate91A majority of Linux being faster by default vs. Windows is the amount of slop and bloat Windows comes with, VBS being on is a Windows problem and it being on by default is fair enough IMO.
Windows 11 is crap once you've tried Linux you see just how crap it really is,Linux being much more smooth to use without bloat, forced updates breaking things, ads, and spyware, although of course ignorance is bliss.
This could have been tested with an LTSC installation, VBS and other things disabled, etc.
Posted on Reply
#17
Grimling
I think it is fair to test Windows 11 as it comes out of the box with everything that is enabled by default and with all the bloat it comes with!
VBS costs performance, yes, but if it is enabled by default, its gonna be enabled on most PCs.
Optimized LTSC versions may be faster, yes, but that is not what is on the average Users PCs.
Posted on Reply
#18
_roman_
I saw similar post hours ago on reddit and partly ignored it

i am not sure, tehy may have backported or reused git related code. To rephrase it, code which is in development etc.

do i want git related code in my linux kernel. nope.

This will hit sooner or later the stable branch anyway. The linux kernel release are very fast compared to other projects.

I do get their point to increase marketing and to increase their speed benefits for the usual microsoft windows user who has no idea

I did bother for years with ricer gcc flags and gave up. You can do it when you want it but there are other means to increase speed and productivity and the memory footprint. get rid of most kde, gnome, qt and gtk apps and of course systemd.

my source: Gentoo/comments/1vp82dp/try_out_the_cachyos_kernel_on_gentoo
  • I only had about a 10% increase in benchmark difference, but was worth it for me.
  • efault kernel (LLVM/Clang Auto FDO Propeller ThinLTO) with eBPF beerland/cake scx scheduler. They mainly help with throughput stability.

    BORE I saw had regression and haven't been using it.

    github.com/CachyOS/linux-cachyos/commit/709f0cf3d9be2d3007d6a0d807d80eb304a203ad

    Looking into core pinning next for gaming to see if it helps with frame drops.
feel Free to dig into those reddit claims and find out if i posted nonsense or if this is a starting point.

Rant: A few have possible xpertise but so many trolls and kids on reddit.

--

I personally miss, not sure if it was stated here, what does cachyos differently?

What are they doing in .config?
what are they copying over?

In the early gentoo years i abused the forums.gentoo.org a lot and got at lesat one ban which i asked to be removed years later. In those years, the performance claims were always below 5%.
The 10% performance claims from cachyos sounds too good to be true. My cooling setup is not optimal on my gentoo box. I did not had the fun yet to replace the cooler and all cpu case fans. Those testing will generate heat which is better done in the cold months so i indirectly will heat my room again.
Posted on Reply
#19
GeoffreyA
GrimlingI think it is fair to test Windows 11 as it comes out of the box with everything that is enabled by default and with all the bloat it comes with!
VBS costs performance, yes, but if it is enabled by default, its gonna be enabled on most PCs.
Optimized LTSC versions may be faster, yes, but that is not what is on the average Users PCs.
There are a few different Linux distributions in the test, one of which is an optimised variant, CachyOS. So, why not test different Windows distributions? Or at least, stock and VBS off. No doubt, even Defender is taking up cycles.
Posted on Reply
#20
Grimling
GeoffreyASo, why not test different Windows distributions?
That is actually a good idea that i support!
On the other hand, after that every Windows-Gamer would want to migrate from home/pro to Enterprise/LTSC...
Posted on Reply
#21
_roman_
ISSEnterpriseIf my memory is correct, best performance isn't Windows' advantage since its early days. The UNIX workstations it replaced still perform better in productivity. But Linux still have very long way to go before becoming a mature desktop operating system for non-programmers.
I could say the same.

I do use my refurbish notebook with official microsoft refurbisher microsoft windows 11 pro on a lenovo l15 gen 1 wiht amd cpu/gpu = apu which is around 5 years old.
The windows explorer crashed yesterday, windows update took forever to reboot, more than 6 minutes, and moving files from my notebook to my tablet caused only two of 3 files being moved. although i waited more than 5 minutes after the windows explorer showed it as completed. the advertisement in the windows 11 pro startmenu, for whatsapp - that was recently added without me wanting too, and other offensive advertisment makes windows not ready as a mature desktop.

Those random crashes i had in past years, 21 years on gentoo but other experience also, came only from testing for uefi values for DRAM.

I accept windows user say always it is not mature, but i use it since 1996 and linux kernel version 2.0.0. I doubt many of those claims come from users who really use both stuff regularly. AND In some extents. i also installed arch linux and had it on my disk for 1 or 2 years from 2006 onwards, same with gentoo. still i did not bootup a lot arch. I had several years of pleasure and blue screens with windows 10 pro on different hardware with microsoft office at work. Display corruption etc. usb not working etc. in windows 10 pro regardless of motherboard and whatever prebuild box it is.

Yes, I agree, windows is not a mature desktop operating system for non-programmers also.

Sheltered in windows 11 pro, on my main box, shows the circular mouse symbol while gaming offline. It seems windows defender cries, makes notification sound without being asked to. the changing of the mouse curser is a bug of windows 11 pro. I mostly use only the monthly windows 11 pro update packages from computerbase.de/downloads for windows 11 pro. the list is very long - why windows is not mature. (keeping cve generic binaries, game binaries and other cve binaries in cve windows 11 pro makes sense - and also smaller backups for my main box)
GrimlingVBS costs performance, yes, but if it is enabled by default, its gonna be enabled on most PCs.
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.
Posted on Reply
#22
AleXXX666
GrimlingOptimized LTSC versions may be faster, yes, but that is not what is on the average Users PCs.
we comparing WORKSTATION PERFORMANCE here, so "Average Joe" or "Ivan" playing CS-GO or WoT doesn't come as good example here. Pro users dealing with such hardware usually have IT staff which configures the software for work, so I don't think there will be "recommendations" here and there popping up when a pro user wants to actually WORK on workstation.:rolleyes::)
GrimlingThat is actually a good idea that i support!
On the other hand, after that every Windows-Gamer would want to migrate from home/pro to Enterprise/LTSC...
SMART 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
Posted on Reply
#23
Grimling
AleXXX666we comparing WORKSTATION PERFORMANCE here, so "Average Joe" or "Ivan" playing CS-GO or WoT doesn't come as good example here. Pro users dealing with such hardware usually have IT staff which configures the software for work, so I don't think there will be "recommendations" here and there popping up when a pro user want to actually WORK on workstation.:rolleyes::)


SMART 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
I remember migrating from 98SE (ignoring ME) to Windows 2000 Pro.
I felt so "PRO" at that time...
Posted on Reply
#24
JustBenching
GrimlingI think it is fair to test Windows 11 as it comes out of the box with everything that is enabled by default and with all the bloat it comes with!
VBS costs performance, yes, but if it is enabled by default, its gonna be enabled on most PCs.
Optimized LTSC versions may be faster, yes, but that is not what is on the average Users PCs.
VBS 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.
Posted on Reply
#25
_roman_
Infy101It also has optional Spectre/Metldown mitigations enabled while the Linux installations only have the default mitigations
WTF?
Same .config which I use. i dropped the few non related hardware changes.

there is also another section about protection features which could be also disabled. I do not have any userspace hypervisor in use on my single user box. The kernel hardening options for sure cost me performance with other options i have enabled.

For a gambling box only you could ditch all of that, even the uefi updates and downgrade the uefi for possible more performance

Posted on Reply
Add your own comment
Nov 13th, 2026 23:04 CST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts