1.\" Copyright (c) 2015 Adrian Chadd <adrian@FreeBSD.org> 2.\" All rights reserved. 3.\" 4.\" Redistribution and use in source and binary forms, with or without 5.\" modification, are permitted provided that the following conditions 6.\" are met: 7.\" 1. Redistributions of source code must retain the above copyright 8.\" notice, this list of conditions and the following disclaimer. 9.\" 2. Redistributions in binary form must reproduce the above copyright 10.\" notice, this list of conditions and the following disclaimer in the 11.\" documentation and/or other materials provided with the distribution. 12.\" 13.\" THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND 14.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 15.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 16.\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE 17.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 18.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 19.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 20.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 21.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 22.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 23.\" SUCH DAMAGE. 24.\" 25.\" $FreeBSD$ 26.\" 27.Dd July 10, 2018 28.Dt NUMA 4 29.Os 30.Sh NAME 31.Nm NUMA 32.Nd Non-Uniform Memory Access 33.Sh SYNOPSIS 34.Cd options SMP 35.Cd options MAXMEMDOM=16 36.Pp 37.In sys/cpuset.h 38.In sys/bus.h 39.Sh DESCRIPTION 40Non-Uniform Memory Access is a computer architecture design which 41involves unequal costs between processors, memory and IO devices 42in a given system. 43.Pp 44In a 45.Nm 46architecture, the latency to access specific memory or IO devices 47depends upon which processor the memory or device is attached to. 48Accessing memory local to a processor is faster than accessing memory 49that is connected to one of the other processors. 50.Pp 51.Nm 52is enabled when the 53.Cd NUMA 54option is used in a kernel configuration 55file and the 56.Cd MAXMEMDOM 57option is set to a value greater than 1. 58.Pp 59Thread and process 60.Nm 61policies are controlled with the 62.Xr cpuset_getdomain 2 63and 64.Xr cpuset_setdomain 2 65syscalls. 66The 67.Xr cpuset 1 68tool is available for starting processes with a non-default 69policy, or to change the policy of an existing thread or process. 70.Pp 71Systems with non-uniform access to I/O devices may mark those devices 72with the local VM domain identifier. 73Drivers can find out their local domain information by calling 74.Xr bus_get_domain 9 . 75.Ss MIB Variables 76The operation of 77.Nm 78is controlled and exposes information with these 79.Xr sysctl 8 80MIB variables: 81.Pp 82.Bl -tag -width indent -compact 83.It Va vm.ndomains 84The number of VM domains which have been detected. 85.Pp 86.It Va vm.phys_locality 87A table indicating the relative cost of each VM domain to each other. 88A value of 10 indicates equal cost. 89A value of -1 means the locality map is not available or no 90locality information is available. 91.Pp 92.It Va vm.phys_segs 93The map of physical memory, grouped by VM domain. 94.El 95.Sh IMPLEMENTATION NOTES 96The current 97.Nm 98implementation is VM-focused. 99The hardware 100.Nm 101domains are mapped into a contiguous, non-sparse 102VM domain space, starting from 0. 103Thus, VM domain information (for example, the domain identifier) is not 104necessarily the same as is found in the hardware specific information. 105Policy information is available in both struct thread and struct proc. 106.Sh SEE ALSO 107.Xr cpuset 1 , 108.Xr cpuset_getaffinity 2 , 109.Xr cpuset_setaffinity 2 , 110.Xr bus_get_domain 9 111.Sh HISTORY 112.Nm 113first appeared in 114.Fx 9.0 115as a first-touch allocation policy with a fail-over to round-robin allocation 116and was not configurable. 117It was then modified in 118.Fx 10.0 119to implement a round-robin allocation policy and was also not configurable. 120.Pp 121The 122.Xr numa_getaffinity 2 123and 124.Xr numa_setaffinity 2 125syscalls and the 126.Xr numactl 1 127tool first appeared in 128.Fx 11.0 129and were removed in 130.Fx 12.0 . 131Current implementation appeared in 132.Fx 12.0 . 133.Pp 134.Sh AUTHORS 135This manual page written by 136.An Adrian Chadd Aq Mt adrian@FreeBSD.org . 137.Sh NOTES 138No statistics are kept to indicate how often 139.Nm 140allocation policies succeed or fail. 141