Preempt-RT时延测试

在安装成功Preempt-RT之后，需要对系统进行了详细的时延测试，以判定该系统是否满足实时性需求。在安装Preempt-RT一文中已经简要介绍了时延的测试方法，本文将进行更为详细的时延测试工作。需要说明的是，短时间的运行cyclictest得出的结果是无意义的，再就是测试系统上没有合适负载的情况下运行cyslictest所得到的延时统计也是无意义的。本文分别在CPU空载和满负载的情况下对Preempt-RT的时延进行测试。

1. 模拟负载

• 安装stress压力测试工具：

sudo apt-get install stress

• 查看自己电脑的核心数：

cd /sys/devices/system/cpu
ls

我的结果显示cpu标号从cpu0~cpu7，因此核心数为8个

• 让CPU满载运行：

strss --cpu 8

• 查看结果：

sudo apt-get install htop
htop

2. 运行测试程序

• 安装绘图工具gnuplot：

sudo apt-get install gnuplot

• 空载测试

1. 不使用stress工具
2. 运行测试指令：

sudo cyclictest -m -Sp90 -i1000 -h400 -D 30m -q >ouput     #为了方便，我只运行了30分钟，实际测试应为1-2天，且不间断运行

3. 运行脚本：

#!/bin/bash
echo "ready for plot \n"
# get maximum latency
max=`grep "Max Latencies" output | tr " " "\n" | sort -n | tail -1 | sed s/^0*//`
echo "max Latency is $max" 
# Grep data lines, remove empty lines and create a common field separator
grep -v -e "^#" -e "^$" output | tr " " "\t" >histogram 
# set the number of cores, for example
cores=8
# Create two-column data sets with latency classes and frequency values for each core
for i in `seq 1 $cores`
do
  column=`expr $i + 1`
  cut -f1,$column histogram >histogram$i
done
# Create plot command header
echo -n "set title \"Latency plot(No load)\"\n\
set terminal png\n\
set xlabel \"Latency (us), max $max us\"\n\
set logscale y\n\
set xrange [0:400]\n\
set yrange [0.8:*]\n\
set ylabel \"Number of latency samples\"\n\
set output \"Noloadplot.png\"\n\
plot " >plotcmd
# Append plot command data references
for i in `seq 1 $cores`
do
  if test $i != 1
  then
    echo -n ", " >>plotcmd
  fi
  cpuno=`expr $i - 1`
  if test $cpuno -lt 10
  then
    title=" CPU$cpuno"
   else
    title="CPU$cpuno"
  fi
  echo -n "\"histogram$i\" using 1:2 title \"$title\" with histeps" >>plotcmd
done
#Execute plot command
gnuplot -persist <plotcmd
# enjoy
firefox plot.png

• 满载测试

1. 使用stress工具

stress --cpu 8  #我的cpu核心数为8个

2. 运行测试指令：

sudo cyclictest -m -Sp90 -i1000 -h400 -D 30m -q >ouput

3. 运行脚本：

#!/bin/bash
echo "ready for plot \n"
# get maximum latency
max=`grep "Max Latencies" output | tr " " "\n" | sort -n | tail -1 | sed s/^0*//`
echo "max Latency is $max" 
# Grep data lines, remove empty lines and create a common field separator
grep -v -e "^#" -e "^$" output | tr " " "\t" >histogram 
# set the number of cores, for example
cores=8
# Create two-column data sets with latency classes and frequency values for each core
for i in `seq 1 $cores`
do
  column=`expr $i + 1`
  cut -f1,$column histogram >histogram$i
done
# Create plot command header
echo -n "set title \"Latency plot(Full load)\"\n\
set terminal png\n\
set xlabel \"Latency (us), max $max us\"\n\
set logscale y\n\
set xrange [0:400]\n\
set yrange [0.8:*]\n\
set ylabel \"Number of latency samples\"\n\
set output \"Fullloadplot.png\"\n\
plot " >plotcmd
# Append plot command data references
for i in `seq 1 $cores`
do
  if test $i != 1
  then
    echo -n ", " >>plotcmd
  fi
  cpuno=`expr $i - 1`
  if test $cpuno -lt 10
  then
    title=" CPU$cpuno"
   else
    title="CPU$cpuno"
  fi
  echo -n "\"histogram$i\" using 1:2 title \"$title\" with histeps" >>plotcmd
done
#Execute plot command
gnuplot -persist <plotcmd
# enjoy
firefox plot.png

上述程序参考自:Create a latency plot from cyclictest histogram data

3. 观察结果

空载时延测试结果：

满载时延测试结果：