[netperf-dev] Result question

[Rick Jones]

I think this one would benefit netperf-talk so I'm going to try
to redirect the discussion there.

De Simone Enzo Javier wrote:
> Hello all,
> 
> sorry by my noob question,  i read the netperf 4.x.x manual, but the 
> print result change and profit not to understand

no worries - the netperf4 manual is still just the netperf2 manual.


There were some issues with proportional fonts and/or tabs in you original
message so I've done a bit of formatting:

>   1   2     3         4        5         6       7     8     9    10     11
> AVE  SET    TEST      DATA    conf     Min     Max    CPU    +/-  SD    +/-
> Over Name   Time      RATE     +/-    Rate    Rate   Util   Util  usec   usec
> Num         sec       mb/s    mb/s    mb/s    mb/s  %/100  %/100  /KB    /KB
> A1   s1     22.03  3587.93  -0.000 3587.93 3587.93 0.5471 -0.0000  2.498 -0.000
> 
> i not understand the columns 1, 2, 5, 9, 10 and 11

Column one I'm not entirely sure about - Stephen could you comment?

Column two - the name of the test set from the config file.

Column five - the width of the 99% (default) confidence
interval for the result in column four.  Only measured on
request.  When measured it will be how far to either side
of the reported result one has to go to be 99% certain that
the "real" average is in that range of the reported average.

Column nine - the same thing for the measured CPU utilization

Column ten - the Service Demand for the test - how many microseconds
of CPU time it took to transfer 1KB of data.

Column 11 - the width of the confidence interval for the service
demand.

Indeed, the output is a bit cryptic, and could benefit from some changes
like emitting the test names of the tests in the set.  We need/want
to update the report generators to be more generic - so there doesn't
need to be one in each test suite.  when that happens (assistance would
be most welcome :) we can make things a bit less cryptic.

happy benchmarking,

rick jones