查看CPU/CACHE的拓扑结构

转自 http://smilejay.com/2017/12/cpu-cache-topology/

Linux上，CPU和Cache相关的拓扑结构，都可以从sysfs文件系统的目录 /sys/devices/system/cpu/ 来获取详细信息。
在网上，找了对CPU相关拓扑结构的解析的两个脚本，觉得还不错；尽管看起来仍有些粗糙，也暂时去改进了。

一个是来自：https://gist.github.com/stedolan/1089968 ； 它可以打印出每个逻辑CPU属于那儿Socket、哪个core，以及与哪些CPU共享L1/L2/L3的Cache。

 

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

#!/bin/bash   unshared () {     grep '^[0-9]\+$' "$1" > /dev/null }   for cpu in $(ls -d /sys/devices/system/cpu/cpu[0-9]* | sort -t u -k 3 -n); do     echo "${cpu##*/}: [Package #$(cat $cpu/topology/physical_package_id), Core #$(cat $cpu/topology/core_id)]"     if ! unshared $cpu/topology/core_siblings_list; then         echo "  same package as $(cat $cpu/topology/core_siblings_list)"     fi     if ! unshared $cpu/topology/thread_siblings_list; then         echo "  same core as    $(cat $cpu/topology/thread_siblings_list)"     fi     for cache in $cpu/cache/index*; do         printf "  %-15s " "L$(cat $cache/level) $(cat $cache/type):"         echo "$(cat $cache/size) $(cat $cache/ways_of_associativity)-way with $(cat $cache/coherency_line_size) byte lines"         if ! unshared $cache/shared_cpu_list; then             printf "  %-15s [%s]\n" "" "shared with $(cat $cache/shared_cpu_list)"         fi     done     echo done

另一个时候cpu_layout.py脚本，来自dpdk项目，用于展示CPU的拓扑结构（不包括cache的信息）：http://dpdk.org/browse/dpdk/tree/usertools/cpu_layout.py

 

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73

#!/usr/bin/env python   # #   BSD LICENSE  （此处省略过多的License信息）   from __future__ import print_function import sys try:     xrange # Python 2 except NameError:     xrange = range # Python 3   sockets = [] cores = [] core_map = {} base_path = "/sys/devices/system/cpu" fd = open("{}/kernel_max".format(base_path)) max_cpus = int(fd.read()) fd.close() for cpu in xrange(max_cpus + 1):     try:         fd = open("{}/cpu{}/topology/core_id".format(base_path, cpu))     except IOError:         continue     except:         break     core = int(fd.read())     fd.close()     fd = open("{}/cpu{}/topology/physical_package_id".format(base_path, cpu))     socket = int(fd.read())     fd.close()     if core not in cores:         cores.append(core)     if socket not in sockets:         sockets.append(socket)     key = (socket, core)     if key not in core_map:         core_map[key] = []     core_map[key].append(cpu)   print(format("=" * (47 + len(base_path)))) print("Core and Socket Information (as reported by '{}')".format(base_path)) print("{}\n".format("=" * (47 + len(base_path)))) print("cores = ", cores) print("sockets = ", sockets) print("")   max_processor_len = len(str(len(cores) * len(sockets) * 2 - 1)) max_thread_count = len(list(core_map.values())[0]) max_core_map_len = (max_processor_len * max_thread_count)  \                       + len(", ") * (max_thread_count - 1) \                       + len('[]') + len('Socket ') max_core_id_len = len(str(max(cores)))   output = " ".ljust(max_core_id_len + len('Core ')) for s in sockets:     output += " Socket %s" % str(s).ljust(max_core_map_len - len('Socket ')) print(output)   output = " ".ljust(max_core_id_len + len('Core ')) for s in sockets:     output += " --------".ljust(max_core_map_len)     output += " " print(output)   for c in cores:     output = "Core %s" % str(c).ljust(max_core_id_len)     for s in sockets:         if (s,c) in core_map:             output += " " + str(core_map[(s, c)]).ljust(max_core_map_len)         else:             output += " " * (max_core_map_len + 1)     print(output)

当NUMA架构下，你可能还需要了解NUMA的分布，包括每个节点上有哪些逻辑处理器、有多少内存等信息，那么使用numactl工具来查看（在CentOS上可以用yum install numactl命令来安装该工具）。

 

1 2 3 4 5 6 7 8 9 10 11 12

[root@jay-linux ~]$ numactl --hardware available: 2 nodes (0-1) node 0 cpus: 0 1 2 3 4 5 6 7 16 17 18 19 20 21 22 23 node 0 size: 64395 MB node 0 free: 62846 MB node 1 cpus: 8 9 10 11 12 13 14 15 24 25 26 27 28 29 30 31 node 1 size: 64507 MB node 1 free: 63676 MB node distances: node   0   1   0:  10  20   1:  20  10

另外，lscpu 命令也是可以查看很多CPU的信息（包括：架构、逻辑CPU数量、核数、主频、Cache大小、NUMA信息等），该命令在CentOS上是在 util-linux 软件包中。

Related Posts:

1. QEMU/KVM中磁盘cache关闭（cache=off/none）
2. Linux查看CPU个数/多核/多线程的查看
3. CentOS 6.x升级Python到2.7
4. shell中的IFS详解