v1.0.4

ktop v1.0.4

Switch from musl static linking to glibc dynamic linking for NVIDIA GPU compatibility.

Changes

• musl binaries can't dlopen glibc-linked shared libraries (like libnvidia-ml.so), causing GPU detection to silently fail on machines with NVIDIA GPUs
• CI now uses cross tool to build inside older-glibc Docker containers, so binaries are both portable and GPU-compatible
• Tested on 4xA100 box where v1.0.3 couldn't detect GPUs

Install / Upgrade

curl -sSfL https://raw.githubusercontent.com/brontoguana/ktop/master/install.sh | bash

Full Changelog: v1.0.3...v1.0.4

v1.0.3

Static linking

Binaries are now statically linked via musl — no more GLIBC version errors on older distros. Single binary, zero dependencies.

Install or upgrade

curl -sSfL https://raw.githubusercontent.com/brontoguana/ktop/master/install.sh | bash

v1.0.2

Changes

• Show total CPU percentage across all cores (e.g. 400% = 4 cores maxed)
• Show available RAM and disk cache size under memory section

Install or upgrade:

curl -sSfL https://raw.githubusercontent.com/brontoguana/ktop/master/install.sh | bash

Full Changelog: v1.0.1...v1.0.2

v1.0.1

Changes

• Removed all Python legacy code (ktop.py, requirements.txt, setup.sh)
• Version now reads from Cargo.toml at compile time
• Cleaned up .gitignore and build instructions

Install

curl -sSfL https://raw.githubusercontent.com/brontoguana/ktop/master/install.sh | bash

Full Changelog: v1.0.0...v1.0.1

ktop v1.0.0 — Rust Rewrite

ktop v1.0.0 — Complete Rust Rewrite

ktop has been completely rewritten in Rust. Same features, dramatically better performance, and now installs as a single static binary with no dependencies.

What's new

• Rewritten in Rust — ~1.2 MB static binary, near-zero CPU usage, ~2-5 MB RAM (down from ~40 MB with Python)
• One-line installer — curl -sSfL https://raw.githubusercontent.com/brontoguana/ktop/master/install.sh | bash
• No dependencies — no Python, no pip, no venv. Just download and run.
• Same features — GPU monitoring (NVIDIA + AMD), CPU sparklines, RAM/swap, network, temps, OOM kills, process tables, 50 color themes
• Instant startup — launches in milliseconds instead of hundreds of milliseconds

Install

curl -sSfL https://raw.githubusercontent.com/brontoguana/ktop/master/install.sh | bash

Or build from source:

git clone https://github.com/brontoguana/ktop.git
cd ktop/ktop-rs
cargo build --release
sudo cp target/release/ktop /usr/local/bin/

Requirements

• Linux (x86_64 or aarch64)
• NVIDIA drivers for NVIDIA GPU monitoring (optional)
• AMD GPUs detected automatically via sysfs

v0.9.0 — AMD GPU Support

AMD GPU Monitoring

ktop now monitors AMD GPUs via Linux sysfs — no new dependencies required. Systems with NVIDIA, AMD, or mixed GPUs show all cards together.

What's new

• AMD GPU support — detected automatically from /sys/class/drm/card*/device/vendor (vendor 0x1002)
• GPU utilization from gpu_busy_percent, VRAM from mem_info_vram_total/mem_info_vram_used
• AMD GPU temperatures from hwmon temp1_input/temp1_crit
• Mixed NVIDIA+AMD systems show all GPUs with unified numbering
• Gracefully handles missing sysfs files (older cards, APUs): util→0%, VRAM→0/0 GB, temp→N/A
• Deduplicates DRM card entries that point to the same PCI device (common on APUs)
• Supports both nvidia-ml-py (preferred) and legacy pynvml for NVIDIA monitoring
• Double-height CPU sparkline (16 levels of precision)
• setup.sh skips tag checkout on feature branches for easier testing

Install

git clone https://github.com/brontoguana/ktop.git && cd ktop && ./setup.sh

v0.8.1

What's New

• OOM tracker now detects systemd-oomd kills in addition to kernel OOM kills
• Fixed capture_output=True + stderr=DEVNULL conflict that silently broke OOM detection
• Added ktop --version flag
• Installer now auto-checkouts the latest tagged release for stable installs

See CHANGELOG.md for full history.