On a dual processor Xeon (EM64T) would you reccomend turning
hypertreading on or off? I tend go for it off dual processor machines
just in case 2 processes end up on the one physical processor rather
than 2 processes on 2 different physical processors. What do you think?<br><br><div><span class="gmail_quote">On 10/9/05, <b class="gmail_sendername">Kevin P. Fleming</b> <<a href="mailto:kpfleming@digium.com">kpfleming@digium.com
</a>> wrote:</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">Eric Bishop wrote:<br>> Have you founy any real life performance benefit of x86_64 (particularly
<br>> EM64T on Xeon) as apposed to plan old x86?<br><br>Yes. On my Athlon-64 2200+ system on my desk, the 'opteron' version<br>encodes a 6722 block sample file in 478ms; the 'i686' version does it in<br>514ms. The 'i386' version is somewhere near 600ms, since it has no fancy
<br>instruction scheduling.<br><br>Interestingly, _all_ of the 32-bit x86 optimized versions run just fine<br>on that machine, meaning that GCC did not opt to use any instructions<br>that are specific to a processor model/family... the performance
<br>improvements come only from scheduling the instruction flow.<br>_______________________________________________<br>--Bandwidth and Colocation sponsored by <a href="http://Easynews.com">Easynews.com</a> --<br><br>Asterisk-Users mailing list
<br><a href="mailto:Asterisk-Users@lists.digium.com">Asterisk-Users@lists.digium.com</a><br><a href="http://lists.digium.com/mailman/listinfo/asterisk-users">http://lists.digium.com/mailman/listinfo/asterisk-users</a><br>
To UNSUBSCRIBE or update options visit:<br> <a href="http://lists.digium.com/mailman/listinfo/asterisk-users">http://lists.digium.com/mailman/listinfo/asterisk-users</a><br></blockquote></div><br>