<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META http-equiv=Content-Type content="text/html; charset=us-ascii">
<STYLE></STYLE>
<META content="MSHTML 6.00.2900.3243" name=GENERATOR></HEAD>
<BODY bgColor=#ffffff background="">
<DIV><FONT face=Arial color=#0000ff size=2><SPAN class=484181710-19022009>Seems
like QueryPerformanceCounter is the most trustful thing for
now.</SPAN></FONT></DIV>
<DIV><FONT face=Arial color=#0000ff size=2><SPAN class=484181710-19022009>One
needs just not to forget that it takes its own time for work.
:(</SPAN></FONT></DIV><BR>
<BLOCKQUOTE dir=ltr
style="PADDING-LEFT: 5px; MARGIN-LEFT: 5px; BORDER-LEFT: #0000ff 2px solid; MARGIN-RIGHT: 0px">
<DIV class=OutlookMessageHeader lang=en-us dir=ltr align=left>
<HR tabIndex=-1>
<FONT face=Tahoma size=2><B>From:</B> Diego Guella
[mailto:diego.guella@sircomtech.com] <BR><B>Sent:</B> Thursday, February 19,
2009 12:00 PM<BR><B>To:</B> Leonid Keller; Fab Tillier; Sean Hefty;
ofw@lists.openfabrics.org<BR><B>Subject:</B> Re: [ofw][patch][WinVerbs tests]
Re: fix IPv6related connectionproblem <BR></FONT><BR></DIV>
<DIV></DIV>
<DIV><FONT face=Arial size=2>----- Original Message ----- <BR>>From:
"Leonid Keller"<BR>>> You can use the __rdtsc ...<BR>>We gave up this
way.<BR>> __rdtsc is complained on the Web to be unstable function,
especially in<BR>>multi-processor environment.<BR><BR>I am coding a test
application right now, and faced the same problem.</FONT></DIV>
<DIV><FONT face=Arial size=2>I haven't found a real solution, and I'm
interested in how you will solve this.</FONT></DIV>
<DIV><FONT face=Arial size=2>These are the informations I
found:<BR></DIV></FONT>
<DIV><FONT face=Arial size=2><A
href="http://www.virtualdub.org/blog/pivot/entry.php?id=106">http://www.virtualdub.org/blog/pivot/entry.php?id=106</A><BR><A
href="http://blogs.msdn.com/oldnewthing/archive/2008/09/08/8931563.aspx">http://blogs.msdn.com/oldnewthing/archive/2008/09/08/8931563.aspx</A><BR><A
href="http://support.microsoft.com/kb/172338/en-us">http://support.microsoft.com/kb/172338/en-us</A><BR><A
href="http://support.microsoft.com/kb/938448/en-us">http://support.microsoft.com/kb/938448/en-us</A><BR><A
href="http://support.microsoft.com/kb/274323/en-us">http://support.microsoft.com/kb/274323/en-us</A><BR></FONT></DIV>
<DIV><FONT face=Arial size=2>I'm still using QueryPerformanceCounter
right now, but I don't really know how much trust to put in
it.<BR></FONT><FONT face=Arial size=2><BR>>If it is that simple why didn't
Microsoft use it in<BR>>QueryPerformanceCounter implementation ?
:(<BR>><BR>>Let's formulate our problem:<BR>>We have to measure a
very fast process and the time T of the<BR>>time stamping itself influence
the results.<BR>><BR>>Possible solutions:<BR>>1. Exclude T from the
result by some artificial trick. (as in the<BR>>discussed patch)<BR>>2.
Measure T before the loop and subtract after.<BR>>3. Make stampings once in
N loops decreasing N times it's influence.<BR>>4. ?<BR>><BR>>What
would you prefere ?<BR>></FONT></DIV></BLOCKQUOTE></BODY></HTML>