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

[Leonid Keller]

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.
The timestamp has been moved to after post, because on Win2k8 it takes
1.1 us !!!
And this is what doubles the latency!
We placed it after post, where the test anyway does nothing...
BTW, the same test with timestamps on your favorite place :)
on win2k3 takes 1.1 usec less, timestamping there takes several nanos.

Another solution would have been a proprietory function, that performs
timestamping on win2k8 fast.

> -----Original Message-----
> From: Sean Hefty [mailto:sean.hefty at intel.com] 
> Sent: Wednesday, February 18, 2009 11:02 PM
> To: 'Fab Tillier'; Leonid Keller; ofw at lists.openfabrics.org
> Subject: RE: [ofw][patch][WinVerbs tests] Re: fix IPv6 
> related connection problem 
> 
> >I would expect that taking the timestamp after the post would 
> >artificially lower the results - the time stamp should be taken just 
> >before posting, so that the time it takes to post is taken 
> into account.
> 
> I agree.  I don't understand moving the time stamp.  Can you 
> please explain?
> 
> >If the IBAL tests take the timestamp after posting, those 
> tests need to 
> >be updated.
> 
> The ibal tests do take the timestamp after posting.
> 
> - Sean
> 
>