The Arch Linux derivative CachyOS has once again demonstrated that distribution choice can have a measurable impact on raw performance, this time on a 64-core workstation that costs as much as a compact car. In a comprehensive benchmark battle on System76’s newly redesigned Thelio Major, CachyOS secured the top spot in 38 out of 104 individual tests, pulling ahead of Arch Linux by roughly 5 percent and Ubuntu 26.04 LTS by about 10 percent in overall geometric mean scores. The results, published by Phoronix, mark the latest in a series of benchmark wins for the performance-tuned distribution, which has now proven its optimizations scale from EPYC servers to high-end desktop workstations.
The Test Platform: A $7,000 Workstation Fight
The battleground was the System76 Thelio Major, a fully redesigned Linux workstation that launched in April 2026. The review unit packed an AMD Ryzen Threadripper 9980X with 64 cores and 128 threads built on the Zen 5 architecture, a Radeon AI PRO R9700 GPU with 32 GB of VRAM, 128 GB of DDR5-5600 ECC memory in a four-stick configuration, and a 1 TB PCIe Gen 5 NVMe SSD from Crucial. The base configuration starts at $6,999 — roughly 1.11 million yen at current exchange rates.
Phoronix installed four distributions clean on this hardware: Ubuntu 26.04 LTS, Pop!_OS 24.04, Arch Linux, and CachyOS. Each was tested with its default settings using the Phoronix Test Suite, which ran 104 separate benchmarks covering everything from video encoding and compiling to compression and cryptographic workloads. The software stacks varied significantly between distributions. CachyOS ran Linux kernel 7.0, Mesa 26.1, and GCC 16.1 — all cutting-edge versions. Arch Linux used similarly recent packages. Ubuntu 26.04 LTS, being a long-term support release, trailed slightly in some components. Pop!_OS 24.04, based on Ubuntu 24.04 LTS, carried packages roughly two years older than what CachyOS and Arch shipped.
CachyOS Took 38 of 104 Tests — Where It Mattered Most
The headline numbers tell a clear story. CachyOS finished first in 38 benchmarks, a 36.5 percent win rate. Arch Linux and Ubuntu 26.04 LTS each won 30 tests, or 28.8 percent. Pop!_OS 24.04 managed only six first-place finishes, a 5.8 percent share.
The geometric mean across all 104 tests — a measure that prevents outlier results from skewing the overall picture — put CachyOS at 38.72, Arch Linux at 36.89, Ubuntu 26.04 LTS at 35.38, and Pop!_OS 24.04 at 31.46. The gap between CachyOS and Pop!_OS exceeded 23 percent. Between CachyOS and Arch, the gap was roughly 5 percent, and between CachyOS and Ubuntu it was about 10 percent.
Video Encoding: Where the Gap Opens Widest
The largest performance deltas appeared in video encoding workloads. In SVT-AV1 4.0 at Preset 5 with 4K input, CachyOS delivered 102.56 frames per second. Arch Linux managed 89.24 fps, and Ubuntu 88.87 fps. That is roughly a 15 percent advantage for CachyOS — a difference that is hard to dismiss as noise when the hardware is identical. At the heavier SVT-AV1 Preset 3 with 4K input, CachyOS achieved 31.75 fps against Arch’s 27.90 fps and Pop!_OS’s 26.78 fps. The trend held consistently: the more demanding the encoding preset, the wider CachyOS’s lead grew.
Not every test went CachyOS’s way. In x265 Bosphorus 1080p encoding, Arch Linux took first place at 124.50 fps, with Ubuntu 26.04 LTS at 121.98 fps and CachyOS at 120.60 fps. The loss shows that performance optimization is not a blanket advantage — it depends on which specific workloads respond best to compiler flags, link-time optimizations, and scheduler tuning. But the overall pattern across 104 tests was unambiguous.
Where Does the Speed Come From
CachyOS’s performance edge is not magic, and it is not the result of exotic patches hidden from the public. The distribution takes a straightforward approach that most experienced Linux users could, in theory, replicate themselves — but only at an impractical cost in time and effort.
Standard distributions like Ubuntu and Arch Linux ship binaries compiled for the x86-64 baseline instruction set, which dates back to AMD64 from 2003. This ensures compatibility across virtually any 64-bit x86 processor. CachyOS abandons that lowest-common-denominator approach. It provides custom repositories with binaries optimized for x86-64-v3, x86-64-v4, and even Zen 4 or Zen 5 specific instruction sets. By assuming the presence of AVX2 and AVX-512, the compiler can generate code that makes much more efficient use of modern CPU pipelines.
The distribution goes further by applying link-time optimization, profile-guided optimization, and BOLT — a binary layout optimization tool developed at Meta — to key packages. The kernel ships with custom-tuned EEVDF scheduler parameters by default, with an optional BORE scheduler for users who want even more aggressive responsiveness tuning. Each individual technique yields a modest gain. Combined across the entire software stack, the improvements compound into the double-digit wins seen in the benchmarks.
What makes CachyOS valuable is not any single optimization but the automation and distribution of the entire pipeline. A knowledgeable Linux user could theoretically rebuild their entire system with these flags. Doing so for every package, maintaining the patches, testing for regressions, and shipping updates — that is the part that is not realistic for an individual.
Pop!_OS: The Most Uncomfortable Number in the Room
The most painful result in these benchmarks belongs to Pop!_OS 24.04. System76 designs and sells the Thelio Major as a premium Linux workstation, and Pop!_OS is the company’s own distribution — the operating system it ships on its own hardware. Yet on the company’s flagship machine, its own OS managed only six first-place finishes out of 104 tests and trailed CachyOS by over 23 percent in geometric mean.
The reason is straightforward. Pop!_OS 24.04 is based on Ubuntu 24.04 LTS, which uses packages that are roughly two years older than what Arch Linux and CachyOS ship. GCC, Mesa, the kernel — all are one to two generations behind. On a cutting-edge machine like the Thelio Major with Zen 5 cores and a Radeon AI PRO GPU, that package age translates directly into lost performance.
System76’s situation is understandable. The company invested heavily in developing COSMIC, its own Rust-based desktop environment, which finally shipped in Pop!_OS 24.04 in December 2025. That is a major strategic bet, and it consumed development resources that could have gone into upstream kernel work or compiler tuning. Prioritizing a desktop experience overhaul over raw benchmark performance is a defensible choice. But the numbers are what they are. On its own workstation, with its own operating system, Pop!_OS is leaving a substantial amount of hardware performance on the table.
What a 10 Percent Performance Swing Means for Professional Users
The broader implication of these results is that the operating system choice can meaningfully affect throughput for professional computing. This contests the long-held assumption in some Linux circles that distribution-level performance differences are negligible — that as long as the kernel and compiler versions are similar, the results will be close.
CachyOS has now disproven that assumption repeatedly. In December 2025, it outperformed Ubuntu 24.04 LTS by roughly 12 percent on an EPYC server. Now it has beaten Ubuntu 26.04 LTS by roughly 10 percent on a 64-core workstation, and Arch Linux by roughly 5 percent. The pattern holds across server, desktop, and workstation form factors. The hardware does not change. The OS and its compilation strategy do.
For professionals who compile code, encode video, or run simulation workloads daily, a 5 to 15 percent speedup is not a theoretical curiosity. It translates into minutes saved per build, frames gained per render, and deadlines met that otherwise would have been missed. A compiler that finishes 5 percent faster over an eight-hour workday frees nearly 25 minutes. An encoder that runs 15 percent faster on a four-hour job cuts over half an hour.
CachyOS is not for everyone. It is a rolling-release distribution based on Arch Linux, and it carries the attendant risks: more frequent updates, potentially less tested package combinations, and a steeper learning curve than Ubuntu LTS or Fedora. But for users who need every bit of performance their hardware can deliver, and who are comfortable managing a rolling system, the numbers make a strong case. Running a general-purpose binary on a Zen 5 processor is, as one commentator put it, like putting regular gasoline into a sports car. It works, but it leaves performance untapped.
The question these benchmarks raise is whether the mainstream distributions will eventually follow CachyOS’s path. Shipping x86-64-v3 or v4 binaries by default would break compatibility with older hardware, which is a nonstarter for general-purpose distributions. But offering performance-tuned repositories as an opt-in, or providing a clear upgrade path for users on modern hardware, would close the gap. Until then, CachyOS remains the benchmark to beat for raw Linux throughput — on servers, on desktops, and now on 64-core workstations. The reference source for all data in this analysis is the Phoronix review.
