About Kernel Documentation Linux Kernel Contact Linux Resources Linux Blog

Documentation / vm / numa


Based on kernel version 4.16.1. Page generated on 2018-04-09 11:53 EST.

1	Started Nov 1999 by Kanoj Sarcar <kanoj@sgi.com>
2	
3	What is NUMA?
4	
5	This question can be answered from a couple of perspectives:  the
6	hardware view and the Linux software view.
7	
8	From the hardware perspective, a NUMA system is a computer platform that
9	comprises multiple components or assemblies each of which may contain 0
10	or more CPUs, local memory, and/or IO buses.  For brevity and to
11	disambiguate the hardware view of these physical components/assemblies
12	from the software abstraction thereof, we'll call the components/assemblies
13	'cells' in this document.
14	
15	Each of the 'cells' may be viewed as an SMP [symmetric multi-processor] subset
16	of the system--although some components necessary for a stand-alone SMP system
17	may not be populated on any given cell.   The cells of the NUMA system are
18	connected together with some sort of system interconnect--e.g., a crossbar or
19	point-to-point link are common types of NUMA system interconnects.  Both of
20	these types of interconnects can be aggregated to create NUMA platforms with
21	cells at multiple distances from other cells.
22	
23	For Linux, the NUMA platforms of interest are primarily what is known as Cache
24	Coherent NUMA or ccNUMA systems.   With ccNUMA systems, all memory is visible
25	to and accessible from any CPU attached to any cell and cache coherency
26	is handled in hardware by the processor caches and/or the system interconnect.
27	
28	Memory access time and effective memory bandwidth varies depending on how far
29	away the cell containing the CPU or IO bus making the memory access is from the
30	cell containing the target memory.  For example, access to memory by CPUs
31	attached to the same cell will experience faster access times and higher
32	bandwidths than accesses to memory on other, remote cells.  NUMA platforms
33	can have cells at multiple remote distances from any given cell.
34	
35	Platform vendors don't build NUMA systems just to make software developers'
36	lives interesting.  Rather, this architecture is a means to provide scalable
37	memory bandwidth.  However, to achieve scalable memory bandwidth, system and
38	application software must arrange for a large majority of the memory references
39	[cache misses] to be to "local" memory--memory on the same cell, if any--or
40	to the closest cell with memory.
41	
42	This leads to the Linux software view of a NUMA system:
43	
44	Linux divides the system's hardware resources into multiple software
45	abstractions called "nodes".  Linux maps the nodes onto the physical cells
46	of the hardware platform, abstracting away some of the details for some
47	architectures.  As with physical cells, software nodes may contain 0 or more
48	CPUs, memory and/or IO buses.  And, again, memory accesses to memory on
49	"closer" nodes--nodes that map to closer cells--will generally experience
50	faster access times and higher effective bandwidth than accesses to more
51	remote cells.
52	
53	For some architectures, such as x86, Linux will "hide" any node representing a
54	physical cell that has no memory attached, and reassign any CPUs attached to
55	that cell to a node representing a cell that does have memory.  Thus, on
56	these architectures, one cannot assume that all CPUs that Linux associates with
57	a given node will see the same local memory access times and bandwidth.
58	
59	In addition, for some architectures, again x86 is an example, Linux supports
60	the emulation of additional nodes.  For NUMA emulation, linux will carve up
61	the existing nodes--or the system memory for non-NUMA platforms--into multiple
62	nodes.  Each emulated node will manage a fraction of the underlying cells'
63	physical memory.  NUMA emluation is useful for testing NUMA kernel and
64	application features on non-NUMA platforms, and as a sort of memory resource
65	management mechanism when used together with cpusets.
66	[see Documentation/cgroup-v1/cpusets.txt]
67	
68	For each node with memory, Linux constructs an independent memory management
69	subsystem, complete with its own free page lists, in-use page lists, usage
70	statistics and locks to mediate access.  In addition, Linux constructs for
71	each memory zone [one or more of DMA, DMA32, NORMAL, HIGH_MEMORY, MOVABLE],
72	an ordered "zonelist".  A zonelist specifies the zones/nodes to visit when a
73	selected zone/node cannot satisfy the allocation request.  This situation,
74	when a zone has no available memory to satisfy a request, is called
75	"overflow" or "fallback".
76	
77	Because some nodes contain multiple zones containing different types of
78	memory, Linux must decide whether to order the zonelists such that allocations
79	fall back to the same zone type on a different node, or to a different zone
80	type on the same node.  This is an important consideration because some zones,
81	such as DMA or DMA32, represent relatively scarce resources.  Linux chooses
82	a default Node ordered zonelist. This means it tries to fallback to other zones
83	from the same node before using remote nodes which are ordered by NUMA distance.
84	
85	By default, Linux will attempt to satisfy memory allocation requests from the
86	node to which the CPU that executes the request is assigned.  Specifically,
87	Linux will attempt to allocate from the first node in the appropriate zonelist
88	for the node where the request originates.  This is called "local allocation."
89	If the "local" node cannot satisfy the request, the kernel will examine other
90	nodes' zones in the selected zonelist looking for the first zone in the list
91	that can satisfy the request.
92	
93	Local allocation will tend to keep subsequent access to the allocated memory
94	"local" to the underlying physical resources and off the system interconnect--
95	as long as the task on whose behalf the kernel allocated some memory does not
96	later migrate away from that memory.  The Linux scheduler is aware of the
97	NUMA topology of the platform--embodied in the "scheduling domains" data
98	structures [see Documentation/scheduler/sched-domains.txt]--and the scheduler
99	attempts to minimize task migration to distant scheduling domains.  However,
100	the scheduler does not take a task's NUMA footprint into account directly.
101	Thus, under sufficient imbalance, tasks can migrate between nodes, remote
102	from their initial node and kernel data structures.
103	
104	System administrators and application designers can restrict a task's migration
105	to improve NUMA locality using various CPU affinity command line interfaces,
106	such as taskset(1) and numactl(1), and program interfaces such as
107	sched_setaffinity(2).  Further, one can modify the kernel's default local
108	allocation behavior using Linux NUMA memory policy.
109	[see Documentation/vm/numa_memory_policy.txt.]
110	
111	System administrators can restrict the CPUs and nodes' memories that a non-
112	privileged user can specify in the scheduling or NUMA commands and functions
113	using control groups and CPUsets.  [see Documentation/cgroup-v1/cpusets.txt]
114	
115	On architectures that do not hide memoryless nodes, Linux will include only
116	zones [nodes] with memory in the zonelists.  This means that for a memoryless
117	node the "local memory node"--the node of the first zone in CPU's node's
118	zonelist--will not be the node itself.  Rather, it will be the node that the
119	kernel selected as the nearest node with memory when it built the zonelists.
120	So, default, local allocations will succeed with the kernel supplying the
121	closest available memory.  This is a consequence of the same mechanism that
122	allows such allocations to fallback to other nearby nodes when a node that
123	does contain memory overflows.
124	
125	Some kernel allocations do not want or cannot tolerate this allocation fallback
126	behavior.  Rather they want to be sure they get memory from the specified node
127	or get notified that the node has no free memory.  This is usually the case when
128	a subsystem allocates per CPU memory resources, for example.
129	
130	A typical model for making such an allocation is to obtain the node id of the
131	node to which the "current CPU" is attached using one of the kernel's
132	numa_node_id() or CPU_to_node() functions and then request memory from only
133	the node id returned.  When such an allocation fails, the requesting subsystem
134	may revert to its own fallback path.  The slab kernel memory allocator is an
135	example of this.  Or, the subsystem may choose to disable or not to enable
136	itself on allocation failure.  The kernel profiling subsystem is an example of
137	this.
138	
139	If the architecture supports--does not hide--memoryless nodes, then CPUs
140	attached to memoryless nodes would always incur the fallback path overhead
141	or some subsystems would fail to initialize if they attempted to allocated
142	memory exclusively from a node without memory.  To support such
143	architectures transparently, kernel subsystems can use the numa_mem_id()
144	or cpu_to_mem() function to locate the "local memory node" for the calling or
145	specified CPU.  Again, this is the same node from which default, local page
146	allocations will be attempted.
Hide Line Numbers


About Kernel Documentation Linux Kernel Contact Linux Resources Linux Blog