/********************************************************************** * Buildroot lmbench使用方法 * 说明： * 想在i.mx6上用Buildroot编译出lmbench，结果发现全是单独的命令，不知道 * 怎么使用，脚本需要用到perl，所以编译需要选好perl。 * * 2018-6-21 深圳 宝安西乡 曾剑锋 *********************************************************************/ 一、参考文档： 1. Performances http://www.armadeus.org/wiki/index.php?title=Performances 2. Linux系统性能测试工具Lmbench https://wiki.deepin.io/mediawiki/index.php?title=Linux%E7%B3%BB%E7%BB%9F%E6%80%A7%E8%83%BD%E6%B5%8B%E8%AF%95%E5%B7%A5%E5%85%B7Lmbench 3. 性能测试工具Lmbench使用说明 https://wenku.baidu.com/view/aa29812727d3240c8447ef9a.html 二、解决办法： 1. 将Buildroot编译出来的文件夹打包，目前我自己的是：/output/build/lmbench-3.0-a9 2. 拷贝到主板上解压； 3. cd /jump/to/lmbench/dir/ 4. cd scripts 5. OS=arm ./config-run （生成的配置放在后面） MULTIPLE COPIES [default 1] 2 Job placement selection: 2 MB [default ] 8 SUBSET (ALL|HARWARE|OS|DEVELOPMENT) [default all] HARDWARE then default value (except for result mailing, choose no) except Processor mhz [default 189 MHz, 5.2910 nanosec clock] 192 FSDIR [default /var/tmp] /mnt/host/ Mail results [default yes] no 6. OS=arm ./results Using config in CONFIG.buildroot Sat Jan 1 00:05:54 UTC 2000 Latency measurements Sat Jan 1 00:06:50 UTC 2000 Calculating file system latency Sat Jan 1 00:06:52 UTC 2000 Local networking Sat Jan 1 00:09:36 UTC 2000 Bandwidth measurements Sat Jan 1 00:17:04 UTC 2000 Calculating context switch overhead Sat Jan 1 00:18:01 UTC 2000 McCalpin's STREAM benchmark Sat Jan 1 00:18:38 UTC 2000 Calculating memory load latency Sat Jan 1 00:27:37 UTC 2000 7. ./getsummary ../results/arm/*， L M B E N C H 3 . 0 S U M M A R Y ------------------------------------ (Alpha software, do not distribute) Basic system parameters ------------------------------------------------------------------------------ Host OS Description Mhz tlb cache mem scal pages line par load bytes --------- ------------- ----------------------- ---- ----- ----- ------ ---- buildroot Linux 4.1.15+ arm 996 64 2 Processor, Processes - times in microseconds - smaller is better ------------------------------------------------------------------------------ Host OS Mhz null null open slct sig sig fork exec sh call I/O stat clos TCP inst hndl proc proc proc --------- ------------- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- buildroot Linux 4.1.15+ 996 0.23 0.56 5.70 12.1 19.7 0.66 4.03 1303 3049 7073 Basic integer operations - times in nanoseconds - smaller is better ------------------------------------------------------------------- Host OS intgr intgr intgr intgr intgr bit add mul div mod --------- ------------- ------ ------ ------ ------ ------ Basic uint64 operations - times in nanoseconds - smaller is better ------------------------------------------------------------------ Host OS int64 int64 int64 int64 int64 bit add mul div mod --------- ------------- ------ ------ ------ ------ ------ Basic float operations - times in nanoseconds - smaller is better ----------------------------------------------------------------- Host OS float float float float add mul div bogo --------- ------------- ------ ------ ------ ------ Basic double operations - times in nanoseconds - smaller is better ------------------------------------------------------------------ Host OS double double double double add mul div bogo --------- ------------- ------ ------ ------ ------ Context switching - times in microseconds - smaller is better ------------------------------------------------------------------------- Host OS 2p/0K 2p/16K 2p/64K 8p/16K 8p/64K 16p/16K 16p/64K ctxsw ctxsw ctxsw ctxsw ctxsw ctxsw ctxsw --------- ------------- ------ ------ ------ ------ ------ ------- ------- buildroot Linux 4.1.15+ 3.6900 10.8 21.9 26.8 46.8 33.3 49.3 *Local* Communication latencies in microseconds - smaller is better --------------------------------------------------------------------- Host OS 2p/0K Pipe AF UDP RPC/ TCP RPC/ TCP ctxsw UNIX UDP TCP conn --------- ------------- ----- ----- ---- ----- ----- ----- ----- ---- buildroot Linux 4.1.15+ 3.690 17.2 26.5 76.4 86.0 *Remote* Communication latencies in microseconds - smaller is better --------------------------------------------------------------------- Host OS UDP RPC/ TCP RPC/ TCP UDP TCP conn --------- ------------- ----- ----- ----- ----- ---- buildroot Linux 4.1.15+ File & VM system latencies in microseconds - smaller is better ------------------------------------------------------------------------------- Host OS 0K File 10K File Mmap Prot Page 100fd Create Delete Create Delete Latency Fault Fault selct --------- ------------- ------ ------ ------ ------ ------- ----- ------- ----- buildroot Linux 4.1.15+ 1998.0 0.207 1.42140 6.259 *Local* Communication bandwidths in MB/s - bigger is better ----------------------------------------------------------------------------- Host OS Pipe AF TCP File Mmap Bcopy Bcopy Mem Mem UNIX reread reread (libc) (hand) read write --------- ------------- ---- ---- ---- ------ ------ ------ ------ ---- ----- buildroot Linux 4.1.15+ 453. 411. 238. 552.7 1214.5 591.4 537.1 1204 596.0 Memory latencies in nanoseconds - smaller is better (WARNING - may not be correct, check graphs) ------------------------------------------------------------------------------ Host OS Mhz L1 $ L2 $ Main mem Rand mem Guesses --------- ------------- --- ---- ---- -------- -------- ------- buildroot Linux 4.1.15+ 996 4.0220 33.7 61.1 316.6 三、生成的配置 DISKS="" DISK_DESC="" OUTPUT=/dev/tty ENOUGH=5000 FASTMEM="NO" FILE=/var/tmp/XXX FSDIR=/var/tmp INFO=INFO.buildroot LINE_SIZE=64 LOOP_O=0.00001841 MAIL=no TOTAL_MEM=1001 MB=8 MHZ="995 MHz, 1.0050 nanosec clock" MOTHERBOARD="" NETWORKS="" OS="arm" PROCESSORS="2" REMOTE="" SLOWFS="NO" SYNC_MAX="2" LMBENCH_SCHED="BALANCED" TIMING_O=0 RSH= RCP= VERSION=3.0-a9 BENCHMARK_HARDWARE=YES BENCHMARK_OS=YES BENCHMARK_SYSCALL= BENCHMARK_SELECT= BENCHMARK_SIG= BENCHMARK_PROC= BENCHMARK_CTX= BENCHMARK_PAGEFAULT= BENCHMARK_FILE= BENCHMARK_MMAP= BENCHMARK_PIPE= BENCHMARK_UNIX= BENCHMARK_UDP= BENCHMARK_TCP= BENCHMARK_CONNECT= BENCHMARK_RPC= BENCHMARK_HTTP= BENCHMARK_BCOPY= BENCHMARK_MEM= BENCHMARK_OPS= 四、getsummary显示数据疑问： 1. 可能前面会发现整形、浮点等运算为什么无效？也就是没有数据。 2. 分析getsummary源代码： [...省略] if (/SYNC_MAX/) { -------+ @_ = split; | $_[1] =~ s/]//; | push(@load, $_[1]); --------+----------------------+ } | [...省略] | for ($i = 0; $i <= $#uname; $i++) { | if (&resultsq($i, $i, ( 'int64_bit', 'int64_add', | 'int64_mul', 'int64_div', | 'int64_mod' )) <= 0) { | next; | } | if ($load[$i] > 1) { <---------------------------------+ next; } printf "%-9.9s %13.13s ", $host[$i], &getos($uname[$i]); printf " %5.5s %6.6s %6.6s %6.6s %6.6s ", &scale_num($int64_bit[$i], 6, $load[$i]), &scale_num($int64_add[$i], 6, $load[$i]), &scale_num($int64_mul[$i], 6, $load[$i]), &scale_num($int64_div[$i], 6, $load[$i]), &scale_num($int64_mod[$i], 6, $load[$i]); } [...省略] 3. 可知SYNC_MAX不能设置大于1的值，否者不会显示相应的部分。