<div dir="ltr">It wasn't that simple (since I stumbled upon numerous complains about incompatible format), so I had to manually set up all compilers to use intel compilers. But OK.<div><br></div><div><br></div><div>Another concern is if pw.x can benefit from setup. Specifically</div><div>- can QE perform dynamic load balancing and utilize heterogeneous nodes efficiently?</div><div>- can QU utilize phi cores efficiently (i.e. if pure native execution would be more efficient than offload or pure CPU modes)?</div></div><div class="gmail_extra"><br><div class="gmail_quote">On 24 March 2015 at 18:19, Fabio Affinito <span dir="ltr"><<a href="mailto:f.affinito@cineca.it" target="_blank">f.affinito@cineca.it</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hi,<br>
<br>
You can take advantage of the offload mode using the libxphi library. This library offloads the BLAS/MKL functions on the Xeon Phi platform hiding the latency times due to the communication. You just need to compile this library and then to link it dynamically. The library works with any version of QE.<br>
Libxphi library is available from <a href="https://github.com/cdahnken/libxphi" target="_blank">https://github.com/cdahnken/libxphi</a><br>
Some documentation is available therein.<br>
<br>
Instead, if you want to compile a native version of pw, you just need to add the -mmic flag and cross compile.<br>
If you want to use the symmetric mode, you need to compile twice: with and without the -mmic flag.<br>
<br>
Best,<br>
<br>
Fabio<br>
<br>
<br>
<br>
----- Messaggio originale -----<br>
> Da: "Евгений Пермяков" <<a href="mailto:permeakra@gmail.com">permeakra@gmail.com</a>><br>
> A: "PWSCF Forum" <<a href="mailto:Pw_forum@pwscf.org">Pw_forum@pwscf.org</a>><br>
> Inviato: Martedì, 24 marzo 2015 15:54:58<br>
> Oggetto: [Pw_forum] compiling for Intel Xeon Phi<br>
><br>
><br>
><br>
> I have access to cluster with Xeon Phi coprocessor on nodes. (it also<br>
> has Scalapack and mkl) The documentaion mentions several way to<br>
> compile programs:<br>
> 1) offload mode, executed on main CPU and offloaded onto coprocessor<br>
> "automagically" (compiled with "mpicc" (I assume, in case of fortran<br>
> applications it is "mpif90")<br>
> 2) native mode, executed completely on coprocessor, compiled with<br>
> "mpicc -mmic" ("mpif90 -mmic"<br>
><br>
> 3) symmetric mode, requiring creation of both binaries<br>
><br>
><br>
> Questions: does pw.x benefit from any of the modes? How should I<br>
> create 'native' binaries ? Is it possible to produce both "mmic" and<br>
> "normal" binaries in same tree or I have to compile to source trees<br>
> separately ?<br>
> _______________________________________________<br>
> Pw_forum mailing list<br>
> <a href="mailto:Pw_forum@pwscf.org">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>
<span class="HOEnZb"><font color="#888888"><br>
--<br>
"When you are solving a problem, don’t worry. Now, after you have solved the problem, then that’s the time to worry." Richard Feynman<br>
<br>
Fabio Affinito, PhD<br>
SuperComputing Applications and Innovation Department<br>
CINECA - via Magnanelli, 6/3, 40033 Casalecchio di Reno (Bologna) - ITALY<br>
Tel: +39 051 6171794 Fax: +39 051 6132198<br>
<br>
_______________________________________________<br>
Pw_forum mailing list<br>
<a href="mailto:Pw_forum@pwscf.org">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></font></span></blockquote></div><br></div>