[netperf-talk] Reported CPU Utilization

[Peter F. Klemperer]

Hi All,

I've seen some interesting results for CPU utilization when running 
netperf that I can't explain.  For message size 1 I see a high CPU 
utilization and then a sharp dropoff with increased message size that 
then increases (see attached graph).  I did not expect to see that 
initial peak and drop.  Can anyone help me understand this?

I have run a series of tests as follows:
./netperf -c -l 10 -H NETSERVIP -w 50 -b 150 -t TCP_STREAM -- -m 1 -D
./netperf -c -l 10 -H NETSERVIP -w 50 -b 150 -t TCP_STREAM -- -m 1024 -D
./netperf -c -l 10 -H NETSERVIP -w 50 -b 150 -t TCP_STREAM -- -m 4098 -D
./netperf -c -l 10 -H NETSERVIP -w 50 -b 150 -t TCP_STREAM -- -m 6144 -D
./netperf -c -l 10 -H NETSERVIP -w 50 -b 150 -t TCP_STREAM -- -m 8192 -D

My test system is a windows virtual machine with 1 CPU, 512 MB RAM and a 
cheapo ethernet card hooked directly to my system running netserver.exe. 
  I then graphed the netperf reported %CPU utilization and saw the trend 
discussed earlier.  To confirm that this is not just an artifact of the 
%CPU implementation I also ran a set of netperf runs in parallel with 
the Dhrystone2.2 benchmark.  Those results are all plotted on the 
attached graph.

Thanks in advance,
Peter F. Klemperer
pklemper at andrew.cmu.edu


-------------- next part --------------
A non-text attachment was scrubbed...
Name: 20110120-cpuover-DHRY-b150-s1000.pdf
Type: application/pdf
Size: 6190 bytes
Desc: not available
URL: <http://www.netperf.org/pipermail/netperf-talk/attachments/20110131/a896cf08/attachment.pdf>