以华为泰山服务器(鲲鹏920芯片)配置为例:
Die Interleaving 控制是否使能DIE交织。使能DIE交织能充分利用系统的DDR带宽,并尽量保证各DDR通道的带宽均衡,提升DDR的利用率
1 for i in $(seq 0 8 24); do echo $i; numactl -C $i -m 0 ./bin/stream -W 5 -N 5 -M 64M; done
2 0
3 STREAM copy latency: 1.22 nanoseconds
4 STREAM copy bandwidth: 13166.34 MB/sec
5 STREAM scale latency: 1.13 nanoseconds
6 STREAM scale bandwidth: 14166.95 MB/sec
7 STREAM add latency: 1.15 nanoseconds
8 STREAM add bandwidth: 20818.63 MB/sec
9 STREAM triad latency: 1.39 nanoseconds
10 STREAM triad bandwidth: 17211.81 MB/sec
11 8
12 STREAM copy latency: 1.56 nanoseconds
13 STREAM copy bandwidth: 10273.07 MB/sec
14 STREAM scale latency: 1.50 nanoseconds
15 STREAM scale bandwidth: 10701.89 MB/sec
16 STREAM add latency: 1.20 nanoseconds
17 STREAM add bandwidth: 19996.68 MB/sec
18 STREAM triad latency: 1.93 nanoseconds
19 STREAM triad bandwidth: 12443.70 MB/sec
20 16
21 STREAM copy latency: 2.52 nanoseconds
22 STREAM copy bandwidth: 6357.71 MB/sec
23 STREAM scale latency: 2.48 nanoseconds
24 STREAM scale bandwidth: 6454.95 MB/sec
25 STREAM add latency: 1.67 nanoseconds
26 STREAM add bandwidth: 14362.51 MB/sec
27 STREAM triad latency: 3.65 nanoseconds
28 STREAM triad bandwidth: 6572.85 MB/sec
29 24
30 STREAM copy latency: 2.44 nanoseconds
31 STREAM copy bandwidth: 6554.24 MB/sec
32 STREAM scale latency: 2.41 nanoseconds
33 STREAM scale bandwidth: 6642.80 MB/sec
34 STREAM add latency: 1.44 nanoseconds
35 STREAM add bandwidth: 16695.82 MB/sec
36 STREAM triad latency: 3.61 nanoseconds
37 STREAM triad bandwidth: 6639.18 MB/sec
38
39 lscpu
40 架构: x86_64
41 CPU 运行模式:32-bit, 64-bit
42 字节序: Little Endian
43 Address sizes: 43 bits physical, 48 bits virtual
44 CPU: 64
45 在线 CPU 列表:0-63
46 每个核的线程数: 2
47 每个座的核数: 16
48 座: 2
49 NUMA 节点:4
50 厂商 ID:HygonGenuine
51 CPU 系列:24
52 型号: 1
53 型号名称: Hygon C86 5280 16-core Processor
54 步进: 1
55 Frequency boost: enabled
56 CPU MHz:2799.311
57 CPU 最大 MHz:2500.0000
58 CPU 最小 MHz:1600.0000
59 BogoMIPS: 4999.36
60 虚拟化: AMD-V
61 L1d 缓存:1 MiB
62 L1i 缓存:2 MiB
63 L2 缓存:16 MiB
64 L3 缓存:64 MiB
65 NUMA 节点0 CPU:0-7,32-39
66 NUMA 节点1 CPU:8-15,40-47
67 NUMA 节点2 CPU:16-23,48-55
68 NUMA 节点3 CPU:24-31,56-63
69 Vulnerability Itlb multihit: Not affected
70 Vulnerability L1tf: Not affected
71 Vulnerability Mds: Not affected
72 Vulnerability Meltdown: Not affected
73 Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl and seccomp
74 Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
75 Vulnerability Spectre v2: Mitigation; Full AMD retpoline, IBPB conditional, STIBP disabled, RSB
76 filling
77 Vulnerability Srbds: Not affected
78 Vulnerability Tsx async abort: Not affected
79 标记: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse3
80 6 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdts
81 cp lm constant_tsc rep_good nopl nonstop_tsc cpuid extd_apicid amd_dcm
82 aperfmperf pni pclmulqdq monitor ssse3 fma cx16 sse4_1 sse4_2 movbe p
83 opcnt xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy
84 abm sse4a misalignsse 3dnowprefetch osvw skinit wdt tce topoext perfct
85 r_core perfctr_nb bpext perfctr_llc mwaitx cpb hw_pstate sme ssbd sev
86 ibpb vmmcall fsgsbase bmi1 avx2 smep bmi2 rdseed adx smap clflushopt s
87 ha_ni xsaveopt xsavec xgetbv1 xsaves clzero irperf xsaveerptr arat npt
88 lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassist
89 s pausefilter pfthreshold avic v_vmsave_vmload vgif overflow_recov suc
90 cor smca
服务热线
1391-024-6332
Copyright 2015-2018 www.intsavi.com.cn All Rights Reserved
电话:010-62980070 010-62961051 手机:13910246332
版权所有北京赛维博信科技发展有限公司 备案号:京ICP备14043711号-1 京ICP备14043711号-3