The machine I decided to work on was Hong Kong, and my package I used to help me time the speed was seahorse.
We created a script would would run and replicate the build processes on the package and provide us a time it finishes a cycle.
Here is the script I used.
------------------------------------------------------------------------------------------------------
#!/bin/bash
rm -f build.log
#Setting the j value, starting from 5 to 15
for ((j=5; j<=15; j++))
do
mv ~/.rpmmacros ~/.rpmmacros.old
#Looks for the smp_flag in the .rpmmacros file and replaces the j value.
grep -v smp_flags ~/.rpmmacros.old >~/.rpmmacros
echo "%_smp_mflags -j$j" >> ~/.rpmmacros
#Run the Build 3 times and Outputting the values to a log
for ((x=1; x<=3; x++))
do
echo "Building...j:$j x:$x"
echo "Time for j value $j Number of Tries: $x" >> build.log
echo "" >>build.log
/usr/bin/time -f "%e %U %s" -o build.log -a rpmbuild -ba /home/bchao/rpmbuild/SPECS/seahorse.spec >/dev/null 2>&1
echo "" >> build.log
done
done
----------------------------------------------------------------------------
Here are my results.
|
| Number of Tries |
|
|
| J Value | 1 | 2 | 3 |
| 5 | 38.55 45.11 0 | 37.93 45.26 0 | 37.99 45.26 0 |
| 6 | 37.00 45.35 0 | 37.46 45.19 0 | 37.25 45.45 0 |
| 7 | 36.80 45.49 0 | 37.20 45.57 0 | 37.15 45.35 0 |
| 8 | 36.26 45.48 0 | 36.04 45.32 0 | 36.37 45.47 0 |
| 9 | 36.03 45.59 0 | 36.00 45.31 0 | 36.23 45.63 0 |
| 10 | 36.32 45.44 0 | 36.02 45.51 0 | 36.11 45.65 0 |
| 11 | 35.88 45.25 0 | 36.04 45.47 0 | 36.20 45.72 0 |
| 12 | 36.42 45.64 0 | 36.01 45.55 0 | 36.10 45.40 0 |
| 13 | 36.31 45.40 0 | 36.12 45.69 0 | 35.98 45.56 0 |
| 14 | 36.12 45.63 0 | 36.62 45.50 0 | 36.01 45.56 0 |
| 15 | 36.03 45.72 0 | 36.11 45.69 0 | 36.14 45.84 0 |
As you can see using the j value 5 gave us the slowest speed of 38 seconds, the fastest time recorded was with the j value 11 at 35.88. The j value where everything basically averages out is 9 with the time of 36 seconds.
So when we do any Fedora builds on Hong Kong, I know which value I would use which is 9.
0 comments:
Post a Comment