lscpu1

Usage

Overview

lscpu gathers CPU architecture information from sysfs, /proc/cpuinfo and any applicable architecture-specific libraries (e.g. librtas on Powerpc). The command output can be optimized for parsing or for easy readability by humans. The information includes, for example, the number of CPUs, threads, cores, sockets, and Non-Uniform Memory Access (NUMA) nodes. There is also information about the CPU caches and cache sharing, family, model, bogoMIPS, byte order, stepping and microcode version.

The default output formatting on a terminal is subject to change and may be optimized for better readability. The output for non-terminals (e.g., pipes) is never affected by this optimization and it is always in "Field: data\n" format. Use for example "lscpu | less" to see the default output without optimizations.

In virtualized environments, the CPU architecture information displayed reflects the configuration of the guest operating system which is typically different from the physical (host) system. On architectures that support retrieving physical topology information, lscpu also displays the number of physical sockets, chips, cores in the host system.

Options that produce an output table accept an optional list argument. This list is a comma-separated series of column labels to limit the table to only the specified columns, arranged in the specified order. Use --help to see a list of valid column labels. The labels are case insensitive. Not all columns are supported on all architectures. If an unsupported column is specified, lscpu prints the column but does not provide any data for it.

The cache sizes are reported as summary from all CPUs. The versions before v2.34 reported per-core sizes, but this output was confusing due to complicated CPUs topology and the way how caches are shared between CPUs. For more details about caches see --cache. Since version v2.37 lscpu follows cache IDs as provided by Linux kernel and it does not always start from zero.

Options

-a, --all

Include lines for online and offline CPUs in the output (default for -e). This option may only be specified together with option -e or -p.

-B, --bytes

Print sizes in bytes rather than in human-readable form.

By default, sizes are shown in units that are powers of 1024 bytes. The formal abbreviations for these units (KiB, MiB, GiB, ...) are further shortened to just their first letter: K, M, G, ....

-b, --online

Limit the output to online CPUs (default for -p). This option may only be specified together with option -e or -p.

-C, --caches[=list]

Display details about CPU caches. For details about available information see --list-columns output.

If the list argument is omitted, all columns for which data is available are included in the command output.

When specifying the list argument, the option string plus the equal sign (=) plus the list may not contain any whitespace. Examples: -C=NAME,ONE-SIZE or --caches=NAME,ONE-SIZE.

The default list of columns may be extended if list is specified in the format +list (e.g., lscpu -C=+ALLOC-POLICY).

-c, --offline

Limit the output to offline CPUs. This option may only be specified together with option -e or -p.

-e, --extended[=list]

Display the CPU information in human-readable format.

If the list argument is omitted, the default columns are included in the command output. The default output is subject to change.

When specifying the list argument, the option string plus the equal sign (=) plus the list may not contain any whitespace. Examples: -e=cpu,node or --extended=cpu,node.

The default list of columns may be extended if list is specified in the format +list (e.g., lscpu -e=+MHZ).

-H, --list-columns

List the columns that can be specified with the --extended, --parse and --caches options. Can be used with --json or --raw to get the list in a machine-readable format.

--hierarchic[=when]

Use subsections in summary output. For backward compatibility, the default is to use subsections only when output on a terminal, and to use flattened output on a non-terminal. The optional argument when can be never, always, or auto. If the when argument is omitted, it defaults to always.

-J, --json

Use JSON output format for the default summary or extended output (see --extended). For backward compatibility, JSON output follows the default summary behavior for non-terminals (e.g., pipes) where subsections are missing. See also --hierarchic.

--output-all

Output all available columns. This option must be combined with either --extended, --parse, or --caches.

-p, --parse[=list]

Optimize the command output for easy parsing.

If the list argument is omitted, the command output is compatible with earlier versions of lscpu. In this compatible format, two commas are used to separate CPU cache columns. If no CPU caches are identified the cache column is omitted. If the list argument is used, cache columns are separated with a colon (:).

When specifying the list argument, the option string plus the equal sign (=) plus the list may not contain any whitespace. Examples: -p=cpu,online,mhz or --parse=cpu,online,mhz.

The default list of columns may be extended if list is specified in the format +list (e.g., lscpu -p=+MHZ).

-r, --raw

Produce output in its raw format. This option is only applicable for the --extended, --parse, and --caches outputs.

-s, --sysroot directory

Gather CPU data for a Linux instance other than the instance from which the lscpu command is issued. The specified directory is the system root of the Linux instance to be inspected.

-x, --hex

Use hexadecimal masks for CPU sets (for example "ff"). The default is to print the sets in list format (for example 0,1). Note that before version 2.30 the mask has been printed with 0x prefix.

-y, --physical

Display physical IDs for all columns with topology elements (core, socket, etc.). Other than logical IDs, which are assigned by lscpu, physical IDs are platform-specific values that are provided by the kernel. Physical IDs are not necessarily unique and they might not be arranged sequentially. If the kernel could not retrieve a physical ID for an element lscpu prints the dash (-) character.

The CPU logical numbers are not affected by this option.

--arm-id[=list]

Print a list of the known ARM core implementers and their human readable names.

If an argument is given, print a list of the individual core IDs and their names, for the given implementer.

--arm-model id

Print the name for this specific core model. This requires passing --arm-id=id to indicate the implementer id as well.

--annotate[=when]

Adds an annotation to each column header name. Such an annotation can be shown as a tooltip by terminals that support this feature. The optional when argument can be always, never, or auto. If the argument is omitted, it defaults to auto, which means that annotations will only be used when the output goes to a terminal.

-h, --help

-V, --version

Environment

LSCPU_COLUMNS=

LSCPU_CACHES_COLUMNS=

LIBSMARTCOLS_DEBUG=all

LIBSMARTCOLS_DEBUG_PADDING=on

LIBSMARTCOLS_JSON=compact|lines

Columns

A list of valid column labels can be viewed with the --list-columns option.

Notes

The default output is subject to change. So whenever possible, you should avoid using default output in your scripts.