[ofw][patch][WinVerbs tests] Re: fix IPv6 related connection problem

[Leonid Keller]

I don't suggest this solution for all times and cases.
Try to run this test on win2003 with time stamp before and after the
post, and you will see no difference.
The time stamping is not a part of the test.
Putting it after the post is just a way to subtract the time T of taking
the time stamp from the roundrobin time.
Another way would be to measure T before the loop and to subtract it
during the result calculation.


> -----Original Message-----
> From: Sean Hefty [mailto:sean.hefty at intel.com] 
> Sent: Thursday, February 19, 2009 1:25 AM
> To: Leonid Keller; Fab Tillier; ofw at lists.openfabrics.org
> Subject: RE: [ofw][patch][WinVerbs tests] Re: fix IPv6 
> related connection problem 
> 
> >After posting the request, the test goes into waiting for the reply 
> >from the other side.
> >This time is much more than the time of posting, so it 
> doesn't in fact 
> >lower the results.
> 
> You can't know this.  After the post, the data could have 
> been sent, received by the other side, and a reply already 
> generated.  In order to time the round trip time, the post 
> must be included.
> 
> Your fix is equivalent to telling runners in a race to go, 
> then starting the stop watch after they're already running.  
> The race might take longer than it takes you to start the 
> stop watch, but the resulting time that you get isn't the 
> amount of time that it took anyone to run the race.
> 
> >Another solution would have been a proprietory function, 
> that performs 
> >timestamping on win2k8 fast.
> 
> How is taking the time stamp the problem?
> 
> - Sean
> 
>