[Pw_forum] Quantum Espresso on Xeon Phi

[Paul.Caheny at uk.fujitsu.com]

Hi,


I'm interested to try to run some of the standard benchmarks for Quantum Espresso (http://qe-forge.org/gf/project/q-e/frs/?action=FrsReleaseBrowse&frs_package_id=36 ) on Xeon Phi, particularly the pwscf-small, GRIR443 and AUSURF112 benchmarks. My background is in HPC, so I only have quite basic knowledge of the QE scientific domain.

I haven't been able to find any information on anyone having run QE on Xeon Phi before on the mailing list or elsewhere, does anyone know of any results for QE on Xeon Phi?

For a start, in trying to run QE natively on a single coprocessor (i.e. all processes/threads within a single Xeon Phi coprocessor), could you suggest some good parallelisation strategies to try out? For example what balance of MPI vs Explicit OpenMP vs Library (MKL) OpenMP and what application specific parallelisation parameters (e.g. -nimage, -npools, -nband, -ntg, -northo or -ndiag)?

The Phi has 60 hardware cores with up to 4 way SMT on each core (and a requirement to use at least 2 of 4 SMT threads per core in order to utilise the hardware fully, as each core will only issue at most an instruction for a single SMT thread every second clock) - therefore it is necessary to use at least 120 (if not 180/240) way concurrency for a single Xeon Phi based run.



Best Regards,

Paul.




---

Paul Caheny
Researcher
Technical Computing Research
Fujitsu Laboratories of Europe Ltd (FLE)
phone: +44 (0) 20 8606 4529
e-mail: Paul.Caheny at uk.fujitsu.com<mailto:Paul.Caheny at uk.fujitsu.com>
web: http://www.fujitsu.com/emea/about/fle/<http://www.fujitsu.com/emea/about/fle>


______________________________________________________________________
                                        
 Fujitsu Laboratories of Europe Limited
 Hayes Park Central, Hayes End Road, Hayes, Middlesex, UB4 8FE
 Registered No. 4153469
 
 This e-mail and any attachments are for the sole use of addressee(s) and
 may contain information which is privileged and confidential. Unauthorised
 use or copying for disclosure is strictly prohibited. The fact that this
 e-mail has been scanned by Trendmicro Interscan does not guarantee that 
 it has not been intercepted or amended nor that it is virus-free.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20131220/86ae4f47/attachment.html>