kernel performance issue

[richard]

Hi:

I have a question about the kernel performance of kernel 2.6.24. I use
"time"
to account the time cost of "date" execution, and find that the "real time"
of kernel 2.6.24 is always bigger than kernel 2.6.21 on my arm board.
For my
understanding, the time cost of 2.6.24 shouldn't be bigger than 2.6.21,
because
it is a upgrade of 2.6.21 and the its performance should be better than
2.6.21,
if I am wrong please let me know.


[root@Linux /root]#cat ./date-loop.sh
#! /bin/sh
loop=1
while [ $loop -le 50 ]
do
         date +%s
         loop=$(($loop+1))
done


Time cost in kernel 2.6.24:
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.48s
user    0m 0.06s
sys     0m 0.43s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.48s
user    0m 0.07s
sys     0m 0.42s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.48s
user    0m 0.09s
sys     0m 0.40s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.48s
user    0m 0.07s
sys     0m 0.41s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.48s
user    0m 0.07s
sys     0m 0.42s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.48s
user    0m 0.09s
sys     0m 0.40s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.48s
user    0m 0.06s
sys     0m 0.43s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.48s
user    0m 0.08s
sys     0m 0.41s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.48s
user    0m 0.09s
sys     0m 0.40s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.48s
user    0m 0.10s
sys     0m 0.39s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.48s
user    0m 0.05s
sys     0m 0.44s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.48s
user    0m 0.05s
sys     0m 0.44s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.48s
user    0m 0.09s
sys     0m 0.40s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.48s
user    0m 0.07s
sys     0m 0.42s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.48s
user    0m 0.07s
sys     0m 0.42s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.48s
user    0m 0.09s
sys     0m 0.40s
[root@Linux /root]#

Time cost in kernel 2.6.21:
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.45s
user    0m 0.08s
sys     0m 0.38s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.45s
user    0m 0.07s
sys     0m 0.39s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.45s
user    0m 0.09s
sys     0m 0.37s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.45s
user    0m 0.09s
sys     0m 0.37s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.45s
user    0m 0.08s
sys     0m 0.38s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.45s
user    0m 0.06s
sys     0m 0.40s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.45s
user    0m 0.09s
sys     0m 0.37s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.45s
user    0m 0.10s
sys     0m 0.36s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.45s
user    0m 0.04s
sys     0m 0.42s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.45s
user    0m 0.09s
sys     0m 0.37s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.45s
user    0m 0.07s
sys     0m 0.39s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.45s
user    0m 0.08s
sys     0m 0.38s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.45s
user    0m 0.10s
sys     0m 0.36s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.45s
user    0m 0.09s
sys     0m 0.37s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.45s
user    0m 0.09s
sys     0m 0.37s
[root@Linux /root]#time ./date-loop.sh > /dev/null
real    0m 0.45s
user    0m 0.07s
sys     0m 0.39s
[root@Linux /root]#

-- 






Best Regards,
Yang Lu

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/