<html><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space; "><br><div><div>On Aug 10, 2009, at 23:07 , Tom Elken wrote:</div><blockquote type="cite"><div>Summary:<br>IBM, SGI and Platform have some comparisons on clusters with "SMT On" of running 1 rank for every core compared to running 2 ranks on every core. In general, on low core-counts, like up to 32 there is about an 8% advantage for running 2 ranks per core. At larger core counts, IBM published a pair of results on 64 cores where the 64-rank performance was equal to the 128-rank performance. Not all of these applications scale linearly, so on some of them you lose efficiency at 128 ranks compared to 64 ranks.<br><br>Details: Results from this year are mostly on Nehalem:<br><a href="http://www.spec.org/mpi2007/results/res2009q3/">http://www.spec.org/mpi2007/results/res2009q3/</a> (IBM)<br><a href="http://www.spec.org/mpi2007/results/res2009q2/">http://www.spec.org/mpi2007/results/res2009q2/</a> (Platform)<br><a href="http://www.spec.org/mpi2007/results/res2009q1/">http://www.spec.org/mpi2007/results/res2009q1/</a> (SGI)<br> (Intel has results with Turbo mode turned on and off<br> in the q2 and q3 results, for a different comparison)<br><br>Or you can pick out the Xeon 'X5570' and 'X5560' results from the list of all results:<br><a href="http://www.spec.org/mpi2007/results/mpi2007.html">http://www.spec.org/mpi2007/results/mpi2007.html</a><br><br>In the result index, when <br>" Compute Threads Enabled" = 2x "Compute Cores Enabled", then you know SMT is turned on.<br>In these cases, you can then check that when <br>" MPI Ranks" = " Compute Threads Enabled" then you are running 2 ranks per core.<br></div></blockquote></div><div><br></div>Tom,<div><br></div><div>Thanks for the neatly compiled information above. I can just add, that I have conducted a fairly detailed analysis of Nehalem compared to HarperTown in my paper <a href="http://www.springerlink.com/content/b34qn674r0m23228/?p=90e0e6dd92594c7b8b49993c7d245ed7&pi=11">An evaluation of Intel’s core i7 architecture using a comparative approach</a> presented at ISC´09. Here, I look at different aspect of the memory hierarchy of the two processors. The benefits from hyperthreading on the said 13 SPEC MPI2007 applications are also studied, although using only a single node, where the advantage is more pronounced</div><div><br></div><div>Thanks,</div><div><br></div><div> </div><div><div apple-content-edited="true"><div><div>Håkon</div><div><br></div></div><br></div><br></div></body></html>