<div>Rick,</div> <div> </div> <div>I got your explaination about CPU Utilization. That seems to be correct.</div> <div> </div> <div>Your suggestion for the "temp" directory also works fine for me.</div> <div> </div> <div>I have not yet tried the histogram output. I get it correctly, then you are suggesting me to modify the makefile and recreate the netserver and netperf.</div> <div>Am I correct?</div> <div> </div> <div>Thanks a lot for all your help</div> <div><BR><BR><B><I>Rick Jones <rick.jones2@hp.com></I></B> wrote:</div> <BLOCKQUOTE class=replbq style="PADDING-LEFT: 5px; MARGIN-LEFT: 5px; BORDER-LEFT: #1010ff 2px solid">anil mishra wrote:<BR>> Rick,<BR>> <BR>> I tried to run the test with -v 3 option. I am running my tests on MP <BR>> system with processors. I used -c and -C both the options. It gives the <BR>> CPU Utilization and interrupt/sec in three columns. ( I am off today, <BR>> tomorrow I will cut and paste
the result here.)<BR>> The problem what I have it gives output something (IIRC) like following:<BR>> <BR>> Total CPU Utilization CPU-1 Utilization CPU-2 Utilization<BR>> <BR>> Total Int/sec Int/sec (for CPU-1) Int/sec (for CPU-2)<BR>> <BR>> <BR>> But when I tried to do the math like:<BR>> <BR>> Total CPU = CPU-1 + CPU-2<BR>> <BR>> This does not seem to match. Rather IIRC, "Total CPU" was somehow ;less <BR>> than "CPU-1". I had the similar observation for the "interrupts/second" <BR>> metric also.<BR><BR>Assume for the moment that each CPU had a total of 100 cycles available <BR>at any one moment. Further assume that on CPU1 60 of those cycles were <BR>used and on CPU2 30 of those cycles were used. The utilization of CPU1 <BR>would be 60%. The CPU utilization of CPU2 would be 30%. However, the <BR>CPU utilization of the system as a whole is not 60%+30% or 90%. The CPU <BR>utilziation of the system as a whole would be
(60+30)/(100+100) or <BR>90/200 or 45%.<BR><BR>Netperf does not report CPU utilization for MP systems as being > 100%.<BR><BR>> <BR>> I will try to add the histogram to see the individual round trip.<BR>> <BR>> For the problem what I see while trying to run the "netserver" on the <BR>> Dell 2900 system, I am not sure what is going on there. The same code <BR>> base seems to be working on the different Windows system. I am not sure <BR>> if anyone elase has seen this issue.<BR><BR>It is assume that c:\\temp exists. From netserver.c:<BR><BR>#ifndef DEBUG_LOG_FILE<BR>#ifndef WIN32<BR>#define DEBUG_LOG_FILE "/tmp/netperf.debug"<BR>#else<BR>#define DEBUG_LOG_FILE "c:\\temp\\netperf.debug"<BR>#endif /* WIN32 */<BR>#endif /* DEBUG_LOG_FILE */<BR><BR><BR>> If I got the Service Demand explaination correctly, it is kind of <BR>> normalization of<BR>> "Service Demand = CPU Utilization / Amont of Data Transfer". Am I <BR>> correct? I will go thru
the documentation to get more ifo about this. <BR><BR>You are correct. It is the quantity of CPU per unit of work.<BR><BR>rick jones<BR></BLOCKQUOTE><BR><p> 
<hr size=1>Yahoo! Messenger with Voice. <a href="http://us.rd.yahoo.com/mail_us/taglines/postman1/*http://us.rd.yahoo.com/evt=39663/*http://voice.yahoo.com">Make PC-to-Phone Calls</a> to the US (and 30+ countries) for 2¢/min or less.