Apple Silicon chips use a hybrid architecture with two types of CPU cores:
P-CPU (Performance Cores)
- High-performance cores optimized for demanding computational tasks
- Higher clock speeds and more powerful execution units
- Consume more power but deliver maximum single-threaded performance
- Used for: Heavy applications, gaming, video editing, intensive calculations
- Typically 4-8 cores depending on chip variant (M1: 4, M2 Pro: 8, etc.)
E-CPU (Efficiency Cores)
- Energy-efficient cores optimized for background tasks and battery life
- Lower clock speeds but much better performance-per-watt
- Designed to handle routine system tasks with minimal power consumption
- Used for: Background processes, web browsing, light multitasking, system maintenance
- Typically 4 cores across most Apple Silicon variants
How FluidTop displays them:
- Individual core utilization shows both types when using
--show_cores - P-cores are typically listed first, followed by E-cores
- Total CPU utilization combines both core types
- Power consumption reflects the hybrid architecture's efficiency benefits
FluidTop supports all Apple Silicon variants:
- M1 family: M1, M1 Pro, M1 Max, M1 Ultra
- M2 family: M2, M2 Pro, M2 Max, M2 Ultra
- M3 family: M3, M3 Pro, M3 Max
- M4 family: M4, M4 Pro, M4 Max (and future variants)
FluidTop requires root access because it uses macOS's powermetrics utility, which accesses low-level hardware performance counters. This is a macOS security requirement, not a FluidTop limitation.
FluidTop uses hardware energy counters provided by powermetrics, which are the same data sources used by macOS's Activity Monitor and system diagnostics. Power readings are highly accurate for relative measurements and trends, though absolute values may vary slightly based on system calibration.
The Neural Engine is Apple's dedicated AI/ML processor designed for machine learning workloads. FluidTop monitors ANE utilization when AI frameworks like Core ML, TensorFlow, or PyTorch offload computations to it. High ANE usage typically indicates active machine learning inference or training.
FluidTop provides more detailed real-time hardware metrics than Activity Monitor, including:
- Individual CPU core utilization (P-cores vs E-cores)
- Real-time power consumption with averaging
- Neural Engine utilization tracking
- Memory usage monitoring
- Terminal-based interface for remote monitoring
This feature is currently in development. For now, FluidTop provides system-wide monitoring. Process-specific monitoring will be available in a future release.
FluidTop automatically adapts to terminal width but currently has fixed height requirements. Ensure your terminal window is large enough to display all monitoring sections. Terminal height adaptation is planned for a future update.
Data export capabilities (CSV, JSON) are planned for a future release. Currently, FluidTop is designed for real-time monitoring only.
Make sure you're running FluidTop with sudo privileges:
sudo fluidtopTry these solutions:
- Resize your terminal window to be larger
- Use a different color theme:
sudo fluidtop --color 0 - Clear your terminal:
clearbefore running FluidTop - Update to the latest version:
uv tool install fluidtop@latest -U
This can happen during long monitoring sessions. Use the --max_count option to automatically restart:
sudo fluidtop --max_count 1000Reduce the refresh rate to lower system impact:
sudo fluidtop --interval 2.0FluidTop is an enhanced and actively maintained fork of asitop with:
- Support for newer Apple Silicon chips (M3, M4+)
- Optimizations for modern terminal emulators (Ghostty compatibility)
- Enhanced AI/ML workload monitoring focus
- Regular updates and bug fixes
- Improved documentation and user experience
Yes, FluidTop can run alongside other monitoring tools like htop, Activity Monitor, or iStat Menus without conflicts. However, running multiple tools simultaneously may slightly increase system resource usage.
Need more help? Check out the main README or submit an issue on GitHub.