Taskset Command for CPU Affinity in Linux

This article provides an overview of the Linux taskset command, detailing how it manages CPU affinity by binding processes to specific processor cores. It covers the core functionality of the command, the concept of processor affinity, practical command syntax for both running and new processes, and the performance benefits of controlling CPU scheduling.

Understanding CPU Affinity

CPU affinity, or processor affinity, is a scheduling feature that binds a process or thread to one or more designated CPU cores. By default, the Linux kernel scheduler dynamically assigns tasks to any available CPU core to balance the overall system workload. However, frequently moving a process between different cores can lead to performance overhead due to CPU cache invalidation. Setting CPU affinity ensures that a process runs exclusively on the assigned cores, preserving cache locality and reducing latency.

The Function of the taskset Command

The taskset command is a standard Linux utility (part of the util-linux package) used to retrieve and manipulate the CPU affinity of a process. It operates either by modifying the CPU mask of an already running process using its Process ID (PID) or by launching a new process with a predefined core assignment.

CPU affinity can be specified in two formats:

• Bitmask: A hexadecimal representation where each bit corresponds to a CPU core (e.g., 0x00000001 for CPU 0, 0x00000003 for CPUs 0 and 1).
• CPU List: A comma-separated or hyphenated list of core numbers using the -c or --cpu-list flag (e.g., 0,2,3 or 0-3), which is more human-readable.

Common Usage and Syntax

Checking the Current Affinity of a Process

To view which cores are assigned to an active process:

taskset -p <PID>

To display the affinity in a readable core list format:

taskset -cp <PID>

Changing the Affinity of a Running Process

To reassign an existing process to specific cores (such as cores 1 and 3):

taskset -cp 1,3 <PID>

Launching a Process with Specific Affinity

To start an application bound strictly to designated cores (for example, cores 0 and 2):

taskset -c 0,2 <command> <arguments>

Key Use Cases and Benefits

• Cache Optimization: Keeping high-throughput tasks on specific cores maximizes L1 and L2 cache hits, significantly reducing memory access latency.
• Workload Isolation: Critical services (such as databases or real-time trading engines) can be pinned to dedicated cores, isolating them from general system processes.
• NUMA Architecture Tuning: On Non-Uniform Memory Access (NUMA) systems, taskset ensures that tasks run on cores directly attached to the local memory node they access most frequently, avoiding interconnect bottlenecks.
• Benchmarking and Testing: Developers use taskset to eliminate scheduling variability when profiling code performance under consistent hardware conditions.