You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: README.md
+19-13Lines changed: 19 additions & 13 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -27,7 +27,7 @@
27
27
28
28
## Overview
29
29
30
-
CachyOS provides a collection of enhanced Linux kernels designed for improved performance across different workloads. Kernels are built from the [CachyOS Linux](https://github.com/CachyOS/linux) fork (currently based on **Linux 7.0**), which merges topic branches on top of upstream stable releases.
30
+
CachyOS provides a collection of enhanced Linux kernels designed for improved performance across different workloads. Kernels are built from the [CachyOS Linux](https://github.com/CachyOS/linux) fork (currently based on **Linux 7.1**), which merges topic branches on top of upstream stable releases.
31
31
32
32
Our kernels feature multiple CPU schedulers, advanced compiler optimizations, and carefully selected patches to deliver better responsiveness and throughput.
33
33
@@ -65,29 +65,33 @@ Each scheduler is optimized for different use cases. We recommend testing each o
65
65
66
66
## Features
67
67
68
-
The CachyOS kernel fork merges topic branches into each release. The current **Linux 7.0** base includes the following integrated branches and enhancements.
68
+
The CachyOS kernel fork merges topic branches into each release. The current **Linux 7.1** base integrates fifteen topic branches (`amd-isp4`, `bbr3`, `cachy`, `cgroup-vram`, `drm-fair`, `fixes`, `hdmi`, `lenovo-wmi`, `mglru`, `preempt-ipi`, `sched-cluster`, `snd-codecs`, `t2`, `vesa-dsc-bpp`, and `vmalloc-free`) on top of upstream stable.
69
69
70
70
### Performance Optimizations
71
71
72
72
-**Advanced Compilation**: Highly customizable PKGBUILD with support for both GCC and Clang compilers
73
-
-**Link Time Optimization (LTO)**: Clang Thin LTO enabled by default on the main `linux-cachyos` package
73
+
-**Link Time Optimization (LTO)**: Clang Thin LTO (v12) enabled by default on the main `linux-cachyos` package
74
74
-**Distributed ThinLTO**: Support for distributed Clang ThinLTO builds to speed up kernel compilation
75
75
-**Profile-Guided Optimization**: AutoFDO + Propeller profiling on the default kernel for optimal code generation ([Learn more](https://cachyos.org/blog/2411-kernel-autofdo/))
76
76
-**Kernel Control Flow Integrity (kCFI)**: Available when using LLVM for enhanced security
77
77
-**Timer Frequency Options**: Configurable between 100Hz, 250Hz, 300Hz, 500Hz, 600Hz, 750Hz, and 1000Hz (default: 1000Hz)
78
78
-**Architecture Optimizations**: Support for x86-64-v3, x86-64-v4, and AMD Zen4 specific builds
79
79
-**Compiler Optimizations**: Advanced GCC flags including `-fivopts` and `-fmodulo-sched`
-**vmalloc Optimizations**: Bulk page freeing in `vfree()`, improved `vrealloc()` shrink/grow paths, and optimized contiguous range freeing
81
82
82
83
### CPU Enhancements
83
84
84
85
-**Multiple Schedulers**: BORE, EEVDF, and BMQ schedulers for different workload optimization
85
-
-**[POC Selector](https://github.com/masahitoS/scx_cake)**: Piece-Of-Cake fast idle CPU selector inspired by scx_cake, reducing wakeup latency
86
+
-**[POC Selector](https://github.com/masahitoS/scx_cake) v2.6.2**: Piece-Of-Cake fast idle CPU selector inspired by scx_cake, reducing wakeup latency
87
+
-**Cluster-Aware Scheduling**: Improved load balancing for CPU clusters, including balancing between CPUs of identical capacity on hybrid architectures
88
+
-**Preempt-IPI Optimizations**: Lower-latency SMP/IPI paths with preemption re-enabled before TLB flushes and streamlined `smp_call_function()` handling
86
89
-**CachyOS Sauce**: Custom `CONFIG_CACHY` configuration with scheduler and system tweaks
87
90
-**AMD P-State Enhancements**: Preferred Core support and latest amd-pstate improvements from linux-next
88
91
-**Real-Time Support**: RT kernel builds available with BORE scheduler integration
89
92
-**Low-Latency Optimizations**: Patches for improved responsiveness and reduced jitter
90
93
-**sched/wait LIFO accept()**: Socket accept() processed in LIFO order for better cache efficiency
94
+
-**Wakeup CPU Affinity**: Prefer the previous CPU for task wakeups to improve cache locality
91
95
92
96
### Networking
93
97
@@ -97,27 +101,29 @@ The CachyOS kernel fork merges topic branches into each release. The current **L
97
101
98
102
-**ZFS Support**: Built-in ZFS filesystem support with pre-compiled modules
99
103
-**NTFS Improvements**: Upstream NTFS driver fixes for MFT mirror validation, attribute bounds checking, and logfile handling
100
-
-**MGLRU Enhancements**: Improved dirty writeback handling, simplified vmscan reclaim statistics, and Cachy Sauce MM tuning (LRU-gen working set protection, compaction/watermark tweaks, hugepage reclaim)
104
+
-**MGLRU Enhancements**: Unified writeback reclaim statistics, simplified vmscan accounting, improved dirty writeback reactivation, and Cachy Sauce MM tuning (LRU-gen working set protection, compaction/watermark tweaks, hugepage reclaim)
105
+
-**Transparent Hugepages**: `ADVISE` mode for shmem/tmpfs transparent hugepages
101
106
-**NVIDIA Integration**:
102
107
- Proprietary NVIDIA driver modules with patches
103
108
- Open-source NVIDIA driver support
104
109
- Ready-to-use modules in repository
105
110
-**I/O Scheduler Improvements**:
106
111
- Enhanced BFQ and mq-deadline performance
107
112
-[ADIOS](https://github.com/firelzrd/adios) v3.2.0 multi-queue I/O scheduler support
108
-
-**VRAM Cgroup (DMEM)**: Device memory controller for restricting GPU VRAM usage per cgroup in the DRM subsystem
113
+
-**VRAM Cgroup (DMEM)**: Device memory controller for restricting GPU VRAM usage per cgroup, with protection limits, common-ancestor pool sharing, and Nouveau DMEM support
109
114
110
115
### Additional Features
111
116
112
117
#### Hardware Support
113
-
-**AMD ISP4**: New AMD ISP 4 camera driver for supported platforms
-**HP Hardware**: OMEN Slim and OMEN MAX laptop support via hp-wmi
119
-
-**Graphics**: HDR support enabled, AMDGPU min_powercap override (`amdgpu_ignore_min_pcap`)
120
-
-**Display**: HDMI VRR on AMD, ALLM and passive VRR connector properties, VESA DSC bits-per-pixel parsing from EDID
121
+
-**ASUS Hardware**: Extended ASUS hardware compatibility patches, including ACP audio fixes for Vivobook 18 and TUF Gaming A14 (2026)
122
+
-**Lenovo Hardware**: WMI battery charge limiting, GPU/CPU tunable attributes, capdata debugfs, and AMD PMC suspend delay for improved sleep/resume
123
+
-**HP Hardware**: OMEN Slim (including 16t-an000) and OMEN MAX laptop support via hp-wmi
124
+
-**Samsung Hardware**: Galaxy Book 4/5 series audio fixes via MAX98390 HDA codec support
125
+
-**Graphics**: HDR support enabled, AMDGPU min_powercap override (`amdgpu_ignore_min_pcap`), and DRM fair GPU scheduling favoring interactive clients
126
+
-**Display**: HDMI VRR on AMD (including PCON path and freesync-on-desktop), ALLM and passive VRR connector properties, VESA DSC bits-per-pixel parsing from EDID
121
127
-**Audio Codecs**: AW88399 and MAX98390 HDA side codec support
0 commit comments