# Testing and monitor the cpu desire performance, frequency, load, # power consumption and throughput etc.when this script trigger tbench # test cases. # 1) Run tbench benchmark on specific governors, ondemand or schedutil. # 2) Run tbench benchmark comparative test on acpi-cpufreq kernel driver. # 3) Get desire performance, frequency, load by perf. # 4) Get power consumption and throughput by amd_pstate_trace.py. # 5) Analyse test results and save it in file selftest.tbench.csv. # 6) Plot png images about performance, energy and performance per watt for each test.
# protect against multiple inclusion if [ $FILE_TBENCH ]; then
return 0 else
FILE_TBENCH=DONE fi
# clear some special lines
clear_csv_tbench()
{ if [ -f $OUTFILE_TBENCH.csv ]; then
sed -i '/Comprison(%)/d' $OUTFILE_TBENCH.csv
sed -i "/$(scaling_name)/d" $OUTFILE_TBENCH.csv fi
}
# find string $1 in file csv and get the number of lines
get_lines_csv_tbench()
{ if [ -f $OUTFILE_TBENCH.csv ]; then
return `grep -c "$1" $OUTFILE_TBENCH.csv` else
return 0 fi
}
grep Joules $OUTFILE_TBENCH-perf-$1-$2.log | awk '{print $4}' > $OUTFILE_TBENCH-energy-$1-$2.log
en_sum=$(awk 'BEGIN {sum=0};{sum += $1};END {print sum}' $OUTFILE_TBENCH-energy-$1-$2.log)
printf "Tbench-$1-#$2 power consumption(J): $en_sum\n" | tee -a $OUTFILE_TBENCH.result
# Permance is throughput per second, denoted T/t, where T is throught rendered in t seconds. # It is well known that P=E/t, where P is power measured in watts(W), E is energy measured in joules(J), # and t is time measured in seconds(s). This means that performance per watt becomes # T/t T/t T # --- = --- = --- # P E/t E # with unit given by MB per joule.
ppw=`echo"scale=4;($TIME_LIMIT-1)*$tp_sum/$en_sum" | bc | awk '{printf "%.4f", $0}'`
printf "Tbench-$1-#$2 performance per watt(MB/J): $ppw\n" | tee -a $OUTFILE_TBENCH.result
printf "\n" | tee -a $OUTFILE_TBENCH.result
# $1: governor
loop_tbench()
{
printf "\nTbench total test times is $LOOP_TIMES for $1\n\n" for i in `seq 1 $LOOP_TIMES` do
run_tbench $1 $i
parse_tbench $1 $i done
}
# $1: governor
gather_tbench()
{
printf "Tbench test result for $1 (loops:$LOOP_TIMES)" | tee -a $OUTFILE_TBENCH.result
printf "\n--------------------------------------------------\n" | tee -a $OUTFILE_TBENCH.result
grep "Tbench-$1-#" $OUTFILE_TBENCH.result | grep "avg des perf:" | awk '{print $NF}' > $OUTFILE_TBENCH-des-perf-$1.log
avg_des_perf=$(awk 'BEGIN {sum=0};{sum += $1};END {print sum/'$LOOP_TIMES'}' $OUTFILE_TBENCH-des-perf-$1.log)
printf "Tbench-$1 avg des perf: $avg_des_perf\n" | tee -a $OUTFILE_TBENCH.result
grep "Tbench-$1-#" $OUTFILE_TBENCH.result | grep "throughput(MB/s):" | awk '{print $NF}' > $OUTFILE_TBENCH-throughput-$1.log
tp_sum=$(awk 'BEGIN {sum=0};{sum += $1};END {print sum}' $OUTFILE_TBENCH-throughput-$1.log)
printf "Tbench-$1 total throughput(MB/s): $tp_sum\n" | tee -a $OUTFILE_TBENCH.result
avg_tp=$(awk 'BEGIN {sum=0};{sum += $1};END {print sum/'$LOOP_TIMES'}' $OUTFILE_TBENCH-throughput-$1.log)
printf "Tbench-$1 avg throughput(MB/s): $avg_tp\n" | tee -a $OUTFILE_TBENCH.result
grep "Tbench-$1-#" $OUTFILE_TBENCH.result | grep "power consumption(J):" | awk '{print $NF}' > $OUTFILE_TBENCH-energy-$1.log
en_sum=$(awk 'BEGIN {sum=0};{sum += $1};END {print sum}' $OUTFILE_TBENCH-energy-$1.log)
printf "Tbench-$1 total power consumption(J): $en_sum\n" | tee -a $OUTFILE_TBENCH.result
avg_en=$(awk 'BEGIN {sum=0};{sum += $1};END {print sum/'$LOOP_TIMES'}' $OUTFILE_TBENCH-energy-$1.log)
printf "Tbench-$1 avg power consumption(J): $avg_en\n" | tee -a $OUTFILE_TBENCH.result
# Permance is throughput per second, denoted T/t, where T is throught rendered in t seconds. # It is well known that P=E/t, where P is power measured in watts(W), E is energy measured in joules(J), # and t is time measured in seconds(s). This means that performance per watt becomes # T/t T/t T # --- = --- = --- # P E/t E # with unit given by MB per joule.
ppw=`echo"scale=4;($TIME_LIMIT-1)*$avg_tp/$avg_en" | bc | awk '{printf "%.4f", $0}'`
printf "Tbench-$1 performance per watt(MB/J): $ppw\n" | tee -a $OUTFILE_TBENCH.result
printf "\n" | tee -a $OUTFILE_TBENCH.result
if [ -n "$base" -a -n "$comp" ]; then
printf "\n==================================================\n" | tee -a $OUTFILE_TBENCH.result
printf "Tbench comparison $1-$2 VS $3-$4" | tee -a $OUTFILE_TBENCH.result
printf "\n==================================================\n" | tee -a $OUTFILE_TBENCH.result
# get the base values
des_perf_base=`echo"$base" | awk '{print $3}' | sed s/,//`
freq_base=`echo"$base" | awk '{print $4}' | sed s/,//`
load_base=`echo"$base" | awk '{print $5}' | sed s/,//`
perf_base=`echo"$base" | awk '{print $6}' | sed s/,//`
energy_base=`echo"$base" | awk '{print $7}' | sed s/,//`
ppw_base=`echo"$base" | awk '{print $8}' | sed s/,//`
# get the comparative values
des_perf_comp=`echo"$comp" | awk '{print $3}' | sed s/,//`
freq_comp=`echo"$comp" | awk '{print $4}' | sed s/,//`
load_comp=`echo"$comp" | awk '{print $5}' | sed s/,//`
perf_comp=`echo"$comp" | awk '{print $6}' | sed s/,//`
energy_comp=`echo"$comp" | awk '{print $7}' | sed s/,//`
ppw_comp=`echo"$comp" | awk '{print $8}' | sed s/,//`
# compare the base and comp values
des_perf_drop=`echo"scale=4;($des_perf_comp-$des_perf_base)*100/$des_perf_base" | bc | awk '{printf "%.4f", $0}'`
printf "Tbench-$1 des perf base: $des_perf_base comprison: $des_perf_comp percent: $des_perf_drop\n" | tee -a $OUTFILE_TBENCH.result
freq_drop=`echo"scale=4;($freq_comp-$freq_base)*100/$freq_base" | bc | awk '{printf "%.4f", $0}'`
printf "Tbench-$1 freq base: $freq_base comprison: $freq_comp percent: $freq_drop\n" | tee -a $OUTFILE_TBENCH.result
load_drop=`echo"scale=4;($load_comp-$load_base)*100/$load_base" | bc | awk '{printf "%.4f", $0}'`
printf "Tbench-$1 load base: $load_base comprison: $load_comp percent: $load_drop\n" | tee -a $OUTFILE_TBENCH.result
perf_drop=`echo"scale=4;($perf_comp-$perf_base)*100/$perf_base" | bc | awk '{printf "%.4f", $0}'`
printf "Tbench-$1 perf base: $perf_base comprison: $perf_comp percent: $perf_drop\n" | tee -a $OUTFILE_TBENCH.result
energy_drop=`echo"scale=4;($energy_comp-$energy_base)*100/$energy_base" | bc | awk '{printf "%.4f", $0}'`
printf "Tbench-$1 energy base: $energy_base comprison: $energy_comp percent: $energy_drop\n" | tee -a $OUTFILE_TBENCH.result
ppw_drop=`echo"scale=4;($ppw_comp-$ppw_base)*100/$ppw_base" | bc | awk '{printf "%.4f", $0}'`
printf "Tbench-$1 performance per watt base: $ppw_base comprison: $ppw_comp percent: $ppw_drop\n" | tee -a $OUTFILE_TBENCH.result
printf "\n" | tee -a $OUTFILE_TBENCH.result
store_csv_tbench "$1-$2 VS $3-$4""Comprison(%)""$des_perf_drop""$freq_drop""$load_drop""$perf_drop""$energy_drop""$ppw_drop" fi
}
# calculate the comparison(%)
calc_comp_tbench()
{ # acpi-cpufreq-ondemand VS acpi-cpufreq-schedutil
__calc_comp_tbench ${all_scaling_names[0]} ${tbench_governors[0]} ${all_scaling_names[0]} ${tbench_governors[1]}
# amd-pstate-ondemand VS amd-pstate-schedutil
__calc_comp_tbench ${all_scaling_names[1]} ${tbench_governors[0]} ${all_scaling_names[1]} ${tbench_governors[1]}
# acpi-cpufreq-ondemand VS amd-pstate-ondemand
__calc_comp_tbench ${all_scaling_names[0]} ${tbench_governors[0]} ${all_scaling_names[1]} ${tbench_governors[0]}
# acpi-cpufreq-schedutil VS amd-pstate-schedutil
__calc_comp_tbench ${all_scaling_names[0]} ${tbench_governors[1]} ${all_scaling_names[1]} ${tbench_governors[1]}
}
# $1: file_name, $2: title, $3: ylable, $4: column
plot_png_tbench()
{ # all_scaling_names[1] all_scaling_names[0] flag # amd-pstate acpi-cpufreq # N N 0 # N Y 1 # Y N 2 # Y Y 3
ret=`grep -c "${all_scaling_names[1]}" $OUTFILE_TBENCH.csv` if [ $ret -eq 0 ]; then
ret=`grep -c "${all_scaling_names[0]}" $OUTFILE_TBENCH.csv` if [ $ret -eq 0 ]; then
flag=0 else
flag=1 fi else
ret=`grep -c "${all_scaling_names[0]}" $OUTFILE_TBENCH.csv` if [ $ret -eq 0 ]; then
flag=2 else
flag=3 fi fi
gnuplot << EOF
set term png
set output "$1"
set title "$2"
set xlabel "Test Cycles (round)"
set ylabel "$3"
set grid
set style data histogram
set style fill solid 0.5 border
set boxwidth 0.8
if ($flag == 1) {
plot \ "<(sed -n -e 's/,//g' -e '/${all_scaling_names[0]}-${tbench_governors[0]}/p' $OUTFILE_TBENCH.csv)" using $4:xtic(2) title "${all_scaling_names[0]}-${tbench_governors[0]}", \ "<(sed -n -e 's/,//g' -e '/${all_scaling_names[0]}-${tbench_governors[1]}/p' $OUTFILE_TBENCH.csv)" using $4:xtic(2) title "${all_scaling_names[0]}-${tbench_governors[1]}"
} else { if ($flag == 2) {
plot \ "<(sed -n -e 's/,//g' -e '/${all_scaling_names[1]}-${tbench_governors[0]}/p' $OUTFILE_TBENCH.csv)" using $4:xtic(2) title "${all_scaling_names[1]}-${tbench_governors[0]}", \ "<(sed -n -e 's/,//g' -e '/${all_scaling_names[1]}-${tbench_governors[1]}/p' $OUTFILE_TBENCH.csv)" using $4:xtic(2) title "${all_scaling_names[1]}-${tbench_governors[1]}"
} else { if ($flag == 3 ) {
plot \ "<(sed -n -e 's/,//g' -e '/${all_scaling_names[0]}-${tbench_governors[0]}/p' $OUTFILE_TBENCH.csv)" using $4:xtic(2) title "${all_scaling_names[0]}-${tbench_governors[0]}", \ "<(sed -n -e 's/,//g' -e '/${all_scaling_names[0]}-${tbench_governors[1]}/p' $OUTFILE_TBENCH.csv)" using $4:xtic(2) title "${all_scaling_names[0]}-${tbench_governors[1]}", \ "<(sed -n -e 's/,//g' -e '/${all_scaling_names[1]}-${tbench_governors[0]}/p' $OUTFILE_TBENCH.csv)" using $4:xtic(2) title "${all_scaling_names[1]}-${tbench_governors[0]}", \ "<(sed -n -e 's/,//g' -e '/${all_scaling_names[1]}-${tbench_governors[1]}/p' $OUTFILE_TBENCH.csv)" using $4:xtic(2) title "${all_scaling_names[1]}-${tbench_governors[1]}"
}
}
}
quit
EOF
}
get_lines_csv_tbench "Governor" if [ $? -eq 0 ]; then # add titles and unit for csv file
store_csv_tbench "Governor""Round""Des-perf""Freq""Load""Performance""Energy""Performance Per Watt"
store_csv_tbench "Unit""""""GHz""""MB/s""J""MB/J" fi
backup_governor for governor in ${tbench_governors[*]} ; do
printf "\nSpecified governor is $governor\n\n"
switch_governor $governor
loop_tbench $governor
gather_tbench $governor done
restore_governor
Die Informationen auf dieser Webseite wurden
nach bestem Wissen sorgfältig zusammengestellt. Es wird jedoch weder Vollständigkeit, noch Richtigkeit,
noch Qualität der bereit gestellten Informationen zugesichert.
Bemerkung:
Die farbliche Syntaxdarstellung und die Messung sind noch experimentell.