numactl - Man Page

• Run a command on node 0 with memory allocated on node 0 and 1: numactl --cpunodebind=0 --membind=0,1 -- command command_arguments
• Run a command on CPUs (cores) 0-4 and 8-12 of the current cpuset: numactl --physcpubind=+0-4,8-12 -- command command_arguments
• Run a command with its memory interleaved on all CPUs: numactl --interleave=all -- command command_arguments

tldr.sh

numactl [ --all ] [ --balancing ] [ --interleave nodes ] [ --preferred node ] [ --preferred-many nodes ] [ --membind nodes ] [ --cpunodebind nodes ] [ --physcpubind cpus ] [ --localalloc ] [--] command {arguments ...}
numactl --show
numactl --hardware [--cpu-compress]
numactl --version
numactl [ --huge ] [ --offset offset ] [ --shmmode shmmode ] [ --length length ] [ --strict ]
[ --shmid id ] --shm shmkeyfile | --file tmpfsfile
[ --touch ] [ --dump ] [ --dump-nodes ] memory policy

numactl runs processes with a specific NUMA scheduling or memory placement policy. The policy is set for command and inherited by all of its children. In addition it can set persistent policy for shared memory segments or files.

Use -- before command if using command options that could be confused with numactl options.

nodes may be specified as N,N,N or  N-N or N,N-N or  N-N,N-N and so forth. Relative nodes may be specified as +N,N,N or  +N-N or +N,N-N and so forth. The + indicates that the node numbers are relative to the process' set of allowed nodes in its current cpuset. A !N-N notation indicates the inverse of N-N, in other words all nodes except N-N.  If used with + notation, specify !+N-N. When same is specified the previous nodemask specified on the command line is used. all means all nodes in the current cpuset.

Instead of a number a node can also be:

Note that block resolves the kernel block device names only for udev names in /dev use file:

Policy settings are:

--all, -a

Unset default cpuset awareness, so user can use all possible CPUs/nodes for following policy settings.

--interleave=nodes, -i nodes

Set a memory interleave policy. Memory will be allocated using round robin on nodes. When memory cannot be allocated on the current interleave target fall back to other nodes. Multiple nodes may be specified on --interleave, --membind and --cpunodebind.

--membind=nodes, -m nodes

Only allocate memory from nodes.  Allocation will fail when there is not enough memory available on these nodes. nodes may be specified as noted above.

--cpunodebind=nodes, -N nodes

Only execute command on the CPUs of nodes. Note that nodes may consist of multiple CPUs. nodes may be specified as noted above.

--physcpubind=cpus, -C cpus

Only execute process on cpus. This accepts cpu numbers as shown in the processor fields of /proc/cpuinfo, or relative cpus as in relative to the current cpuset. You may specify "all", which means all cpus in the current cpuset. Physical cpus may be specified as N,N,N or  N-N or N,N-N or  N-N,N-N and so forth. Relative cpus may be specified as +N,N,N or  +N-N or +N,N-N and so forth. The + indicates that the cpu numbers are relative to the process' set of allowed cpus in its current cpuset. A !N-N notation indicates the inverse of N-N, in other words all cpus except N-N.  If used with + notation, specify !+N-N.

--localalloc, -l

Try to allocate on the current node of the process, but if memory cannot be allocated there fall back to other nodes.

--preferred=node, -p node

Preferably allocate memory on node, but if memory cannot be allocated there fall back to other nodes. This option takes only a single node number. Relative notation may be used.

--balancing, -b

Enable Linux kernel NUMA balancing for the process if it is supported by kernel. This should only be used with --membind, -m only, otherwise ignored.

--preferred-many=nodes, -P nodes

Preferably allocate memory on nodes, but if memory cannot be allocated there fall back to other nodes. This option takes a mask of preferred nodes where the closest node to local is considered most preferred. Relative notation may be used.

--show, -s

Show NUMA policy settings of the current process.

--hardware, -H

Show inventory of available nodes on the system. When the --cpu-compress option is set show cpu ranges. This is not default not break any existing scripts.

Numactl can set up policy for a SYSV shared memory segment or a file in shmfs/hugetlbfs.

--version

print the version of the numactl package and exit.

The following policy settings are persistent and will be used by all mappings from that shared memory. The order of options matters here. The specification must at least include either of --shm, --shmid, --file to specify the shared memory segment or file and a memory policy like described  above ( --interleave, --localalloc, --preferred, --preferred-many, --membind ).

--huge

When creating a SYSV shared memory segment use huge pages. Only valid before --shmid or --shm

--offset

Specify offset into the shared memory segment. Default 0.  Valid units are m (for MB), g (for GB), k (for KB), otherwise it specifies bytes.

--strict

Give an error when a page in the policied area in the shared memory segment already was faulted in with a conflicting policy. Default is to silently ignore this.

--shmmode shmmode

Only valid before --shmid or --shm When creating a shared memory segment set it to numeric mode shmmode.

--length length

Apply policy to length range in the shared memory segment or make  the segment length long Default is to use the remaining length  Required when a shared memory segment is created and specifies the length of the new segment then. Valid units are m (for MB), g (for GB), k (for KB), otherwise it specifies bytes.

--shmid id

Create or use a shared memory segment with numeric ID id

--shm shmkeyfile

Create or use a shared memory segment, with the ID generated using ftok(3) from shmkeyfile

--file tmpfsfile

Set policy for a file in tmpfs or hugetlbfs

--touch

Touch pages to enforce policy early. Default is to not touch them, the policy is applied when an applications maps and accesses a page.

--dump

Dump policy in the specified range.

--dump-nodes

Dump all nodes of the specific range (very verbose!)

Valid node specifiers

all	All nodes
number	Node number
number1{,number2}	Node number1 and Node number2
number1-number2	Nodes from number1 to number2
! nodes	Invert selection of the following specification.

numactl --physcpubind=+0-4,8-12 myapplic arguments Run myapplic on cpus 0-4 and 8-12 of the current cpuset.

numactl --interleave=all bigdatabase arguments Run big database with its memory interleaved on all CPUs.

numactl --cpunodebind=0 --membind=0,1 process Run process on node 0 with memory allocated on node 0 and 1.

numactl --cpunodebind=0 --membind=0,1 -- process -l Run process as above, but with an option (-l) that would be confused with a numactl option.

numactl --cpunodebind=0 --balancing --membind=0,1 process Run process on node 0 with memory allocated on node 0 and 1.  Optimize the page placement with Linux kernel NUMA balancing mechanism if possible.

numactl --cpunodebind=netdev:eth0 --membind=netdev:eth0 network-server Run network-server on the node of network device eth0 with its memory also in the same node.

numactl --preferred=1 numactl --show Set preferred node 1 and show the resulting state.

numactl --preferred-many=0x3 numactl --show Set preferred nodes 1 and 2, and show the resulting state.

numactl --length 1g --shm /tmp/shmkey --interleave=all Interleave all of the sysv shared memory region of size 1g specified by /tmp/shmkey over all nodes.

Place a tmpfs file on 2 nodes:
 numactl --membind=2 dd if=/dev/zero of=/dev/shm/A bs=1M count=1024
 numactl --membind=3 dd if=/dev/zero of=/dev/shm/A seek=1024 bs=1M count=1024

numactl --localalloc /dev/shm/file Reset the policy for the shared memory file file to the default localalloc policy.

Requires a NUMA policy aware kernel.

Command is not executed using a shell. If you want to use shell metacharacters in the child use sh -c as wrapper.

Setting policy for a hugetlbfs file does currently not work because it cannot be extended by truncate.

Shared memory segments larger than numactl's address space cannot  be completely policied. This could be a problem on 32bit architectures. Changing it piece by piece may work.

The old --cpubind which accepts node numbers, not cpu numbers, is deprecated and replaced with the new --cpunodebind and --physcpubind options.

/proc/cpuinfo for the listing of active CPUs. See proc(5) for details.

/sys/devices/system/node/node*/numastat for NUMA memory hit statistics.

Copyright 2002,2004 Andi Kleen, SuSE Labs. numactl and the demo programs are under the GNU General Public License, v.2

set_mempolicy(2) , get_mempolicy(2) , mbind(2) , sched_setaffinity(2) , sched_getaffinity(2) , proc(5) , ftok(3) , shmat(2) , migratepages(8)

autohbw(7), cpuset(7), cyclictest(8), get_mempolicy(2), hbwallocator(3), hbwmalloc(3), mbind(2), memhog(8), memkind(3), memkind-auto-dax-kmem-nodes(1), memkind-hbw-nodes(1), migratepages(8), migspeed(8), numa(3), numa(7), numad(8), numastat(8), sched_setaffinity(2), set_mempolicy(2).