<div dir="ltr"><div>Dear Fabio Affinito</div><div><br></div>Thank you so much for information. </div><div class="gmail_extra"><br clear="all"><div>"Apologizing does not mean that you are wrong and the other one is right...<br>
It simply means that you value the relationship much more than your ego.." </div>
<br><br><div class="gmail_quote">On Wed, Jul 23, 2014 at 5:32 PM, Nisha Agrawal <span dir="ltr"><<a href="mailto:itlinkstonisha@gmail.com" target="_blank">itlinkstonisha@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div dir="ltr">Hi, <div><br></div><div>I setup the quantum espresso Intel Xeon Phi version using the instruction provided in the following link</div><div> </div><div><a href="https://software.intel.com/en-us/articles/quantum-espresso-for-intel-xeon-phi-coprocessor" target="_blank">https://software.intel.com/en-us/articles/quantum-espresso-for-intel-xeon-phi-coprocessor</a> </div>
<div><br></div><div>However when I was running, its not getting offloaded to Intel Xeon PHI , following is the script I am using</div><div>to run QE MIC version. Please let me know If I missed something which is required to set or doing somthing</div>
<div>wrong.</div><div><br></div><div>-------------------------------------------------------------------------------------------------------------------------------------------------------</div><div><div>source /home/opt/ICS-2013.1.039-intel64/bin/compilervars.sh intel64</div>
<div>source /home/opt/ICS-2013.1.039-intel64/mkl/bin/mklvars.sh intel64</div><div>source /home/opt/ICS-2013.1.039-intel64/impi/<a href="http://4.1.2.040/bin64/mpivars.sh" target="_blank">4.1.2.040/bin64/mpivars.sh</a></div>
</div><div><br>
</div><div><div><div>export MKL_MIC_ENABLE=1</div><div>export MKL_DYNAMIC=false</div><div>export MKL_MIC_DISABLE_HOST_FALLBACK=1</div><div>export MIC_LD_LIBRARY_PATH=$MKLROOT/lib/mic:$MIC_LD_LIBRARY_PATH</div></div></div>
<div><br></div><div>export OFFLOAD_DEVICES=0<br></div><div><br></div><div><div>export I_MPI_FALLBACK_DEVICE=disable</div><div>export I_MPI_PIN=disable</div><div>export I_MPI_DEBUG=5</div><div><br></div><div><br></div><div>
export MKL_MIC_ZGEMM_AA_M_MIN=500</div><div>export MKL_MIC_ZGEMM_AA_N_MIN=500</div><div>export MKL_MIC_ZGEMM_AA_K_MIN=500</div><div>export MKL_MIC_THRESHOLDS_ZGEMM=500,500,500</div><div><br></div><div><br></div><div>export OFFLOAD_REPORT=2</div>
</div><div>mpirun -np 8 -perhost 4 ./espresso-5.0.2/bin/pw.x -in ./BN.in 2>&1 | tee test.log<br></div><div><br></div><div>---------------------------------------------------------------------</div><div>------------------------------------------------------------------------------- </div>
<div><br></div><div class="gmail_extra"><br clear="all"><div>"Apologizing does not mean that you are wrong and the other one is right...<br>It simply means that you value the relationship much more than your ego.." </div>
<br><br><div class="gmail_quote">On Mon, Jul 14, 2014 at 8:16 PM, Axel Kohlmeyer <span dir="ltr"><<a href="mailto:akohlmey@gmail.com" target="_blank">akohlmey@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div>On Mon, Jul 14, 2014 at 9:34 AM, Eduardo Menendez <<a href="mailto:eariel99@gmail.com" target="_blank">eariel99@gmail.com</a>> wrote:<br>
> Thank you Axel. Your advise rises another doubt. Can we get the maximum<br>
> performance from a highly clocked CPU?<br>
> I used to consider that the the fastest CPUs were too fast for the memory<br>
> access, resulting in bottlenecks. Of couse it depends on cache size.<br>
<br>
</div>your concern is justified, but the situation is more complex these<br>
days. highly clocked CPUs have less cores and thus receive a larger<br>
share of the available memory bandwidth and the highest clocked<br>
inter-CPU and memory bus is only available for a subset of the CPUs.<br>
now you have an optimization problem that has to consider the strong<br>
scaling (or lack thereof) of the code in question as an additional<br>
input parameter.<br>
<br>
to give an example: we purchased at the same time dual socket nodes<br>
that had the same mainboard, but either 2x 3.5GHz quad-core or 2x<br>
2.8GHz hex-core. the 3.5GHz was the fastest clock available at the<br>
time. for classical MD, i get better performance out of the 12-core<br>
nodes, for plane-wave DFT i get about the same performance out of<br>
both, for CP2k i get better performance with the 8-core (in fact, CP2k<br>
runs fastest on the 12-core with using only 8 cores). now, the cost of<br>
the 2.8GHz CPUs is significantly lower, so that is why we procured the<br>
majority of the cluster with those. but we do have applications that<br>
scale less than CP2k or are serial, but require high per-core memory<br>
bandwidth, so we got a few of the 3.5GHz ones, too (and since they are<br>
already expensive we filled them with RAM as much as it doesn't result<br>
in underclocking of the memory bus; and in turn we put "only" 1GB/core<br>
into the 12-core nodes).<br>
<br>
so it all boils down to finding the right balance and adjusting it to<br>
the application mix that you are running. last time i checked the<br>
intel spec sheets, it looked as if the best deal was to be had for<br>
CPUs with the second largest number of CPU cores and as high a clock<br>
as required to have the full memory bus speed. that will also keep the<br>
heat in check, as the highest clocked CPUs usually have a much higher<br>
TDP (>50% more) and that is just a much larger demand on cooling and<br>
power and will incur additional indirect costs as well.<br>
<br>
HTH,<br>
axel.<br>
<div><br>
<br>
><br>
>>Stick with the cpu. For QE you should be best off with intel. Also you are<br>
>> likely to >get the best price/performance ratio with CPUs that have less<br>
>> than the maximum >number of cpu cores and a higher clock instead.<br>
><br>
><br>
> Eduardo Menendez Proupin<br>
> Departamento de Fisica, Facultad de Ciencias, Universidad de Chile<br>
> URL: <a href="http://www.gnm.cl/emenendez" target="_blank">http://www.gnm.cl/emenendez</a><br>
><br>
> “Science may be described as the art of systematic oversimplification.” Karl<br>
> Popper<br>
><br>
><br>
</div><div>> _______________________________________________<br>
> Pw_forum mailing list<br>
> <a href="mailto:Pw_forum@pwscf.org" target="_blank">Pw_forum@pwscf.org</a><br>
> <a href="http://pwscf.org/mailman/listinfo/pw_forum" target="_blank">http://pwscf.org/mailman/listinfo/pw_forum</a><br>
<br>
<br><span class="HOEnZb"><font color="#888888">
<br>
</font></span></div><span class="HOEnZb"><font color="#888888"><span><font color="#888888">--<br>
Dr. Axel Kohlmeyer <a href="mailto:akohlmey@gmail.com" target="_blank">akohlmey@gmail.com</a> <a href="http://goo.gl/1wk0" target="_blank">http://goo.gl/1wk0</a><br>
College of Science & Technology, Temple University, Philadelphia PA, USA<br>
International Centre for Theoretical Physics, Trieste. Italy.<br>
</font></span><div><div><br>
_______________________________________________<br>
Pw_forum mailing list<br>
<a href="mailto:Pw_forum@pwscf.org" target="_blank">Pw_forum@pwscf.org</a><br>
<a href="http://pwscf.org/mailman/listinfo/pw_forum" target="_blank">http://pwscf.org/mailman/listinfo/pw_forum</a></div></div></font></span></blockquote></div><br></div></div>
</blockquote></div><br></div>