How to use the command numactl (with examples)

Numactl is a command-line utility that allows users to control the NUMA (Non-Uniform Memory Access) policy for processes or shared memory on Linux systems. NUMA is a computer architecture design where a system has multiple memory nodes, and each node is directly connected to a subset of processors (CPUs). This architecture is beneficial for scalability and memory access, especially in large-scale systems.

Use case 1: Run a command on node 0 with memory allocated on node 0 and 1

Code:

numactl --cpunodebind=0 --membind=0,1 -- command command_arguments

Motivation: The motivation behind this use case is to bind a command to a specific CPU node (node 0) and allocate memory from multiple CPU nodes (node 0 and node 1). By doing so, we can optimize the performance by leveraging the locality of data access.

Explanation:

• --cpunodebind=0: This option binds the command to run only on CPUs of node 0.
• --membind=0,1: This option allocates memory from both CPU nodes 0 and 1.
• -- command command_arguments: This is the actual command that we want to run with the specified NUMA policy.

Example output: Assuming we are running a command my_command that requires memory access from multiple nodes, the output could be as follows:

Node: 0 | Command: my_command with arguments
Node: 1 | Command: my_command with arguments

Use case 2: Run a command on CPUs 0-4 and 8-12 of the current cpuset

Code:

numactl --physcpubind=+0-4,8-12 -- command command_arguments

Motivation: The motivation behind this use case is to restrict a command to run on specific CPUs. By doing so, we can control the affinity of the command to certain CPU cores for better performance and resource utilization.

Explanation:

• --physcpubind=+0-4,8-12: This option binds the command to run only on the specified set of CPUs.
  • + sign indicates that the specified CPUs are added to the existing cpuset.
  • 0-4 represents CPUs 0 to 4.
  • 8-12 represents CPUs 8 to 12.
• -- command command_arguments: This is the actual command that we want to run with the specified NUMA policy.

Example output: Assuming we are running a command my_command that performs CPU-intensive tasks, the output could be as follows:

Running my_command on CPUs: 0, 1, 2, 3, 4, 8, 9, 10, 11, 12

Use case 3: Run a command with its memory interleaved on all CPUs

Code:

numactl --interleave=all -- command command_arguments

Motivation: The motivation behind this use case is to distribute the memory interleaving evenly across all available CPUs. By using this approach, we can achieve a balanced memory access pattern, improving overall system performance and avoiding potential bottlenecks.

Explanation:

• --interleave=all: This option enables memory interleaving on all available CPUs.
• -- command command_arguments: This is the actual command that we want to run with the specified NUMA policy.

Example output: Assuming we are running a command my_command that heavily relies on memory access, the output could be as follows:

Running my_command with interleaved memory on all available CPUs.

Conclusion:

The numactl command is a powerful utility for controlling the NUMA policy of processes or shared memory. By leveraging the different options provided by numactl, users can optimize resource allocation, improve performance, and exploit the benefits of NUMA architecture in their Linux systems.