What Is the PRI Column in htop?

This article provides a quick overview of the “PRI” column in the htop Linux command-line tool, explaining its core meaning, how it relates to process scheduling, and how it interacts with the “NI” (Nice) value. You will learn how Linux determines process priority and how to interpret these numbers when monitoring system performance.

Understanding PRI in htop

In the htop interface, PRI stands for Priority. It represents the internal scheduling priority assigned to a process by the Linux kernel. Essentially, the PRI value tells you how much urgency the kernel’s scheduler gives to a particular process when distributing CPU time. A lower PRI value means a higher priority, meaning the process gets CPU attention faster than processes with higher numbers.

How PRI and NI (Nice Value) Work Together

The PRI column is deeply interconnected with the NI column, which stands for Nice value.

• Nice Value (NI): This is a user-controlled value that ranges from -20 (highest priority) to 19 (lowest priority). A default, unaltered process starts with a Nice value of 0.
• Kernel Priority (PRI): The kernel uses the Nice value to calculate the final PRI.

For standard user processes (often called non-real-time processes), the Linux kernel typically calculates the PRI using a simple mathematical baseline:

$$ = + 20$$

Because of this formula, a standard process with a default Nice value of 0 will show a PRI of 20 in htop. If you “renice” a process to make it less urgent—say, an NI of 10—its PRI value will adjust to 30. Conversely, if you grant a process maximum urgency with an NI of -20, its PRI will drop to 0.

Real-Time Priorities in htop

When looking at htop, you might occasionally see negative numbers in the PRI column or the literal text RT.

• RT / Negative Numbers: These indicate Real-Time processes. Real-time processes operate outside the standard 0 to 39 PRI range used by regular applications.
• System Crucial Tasks: The kernel schedules these processes with absolute preference to ensure critical system tasks (like audio processing or hardware drivers) face zero delays, regardless of how busy the CPU is with regular user software.