[Difx-users] IPP Licensing

[Adam Deller]

Hi Stuart,

Peter gave a nice summary of the IPP situation so let me just comment on
the other aspect of your question - a non-IPP compilation of DiFX.  If you
set --disable-ipp when running ./configure for mpifxcorr (or --noipp when
running install-difx) then it will compile using generic vector
instructions (for loops, etc, plus fftw for the FFT).  Note that there are
some (non-essential) other components than mpifxcorr that also depend on
IPP, but they don't have the generic option so they simply won't be built
if no IPP is available.

There has been a bit of a side project to implement some of those generics
with AVX instructions, etc, to speed things up but it hasn't had lots of
attention because IPP just works well for mpifxcorr.  On a "normal" Intel
processor you'll currently experience quite some slowdown if you use the
generic option.

Cheers,
Adam

On Fri, May 6, 2016 at 7:40 AM, Stuart Weston <nzobservers at gmail.com> wrote:

> Just wondering what people do for licensing IPP on a cluster/blade server ?
>
> I have two multi cpu/core servers and my IPP V8 key worked fine on those.
> But it is not working for a node in a blade server ? Also finding how to
> get another "free" academic license for the old version V8 on the Intel
> website is not very intuitive :-( It's quite happy giving me a new license
> for the latest full studio version which includes IPP V9.
>
> More general any plans to move DiFX away from IPP ? I am no expert so not
> sure what else is available or could be used in its stead.
>
>
> _______________________________________________
> Difx-users mailing list
> Difx-users at listmgr.nrao.edu
> https://listmgr.nrao.edu/mailman/listinfo/difx-users
>
>


-- 
!=============================================================!
Dr. Adam Deller
Ph  +31 521595785 / Fax +31 521595101
Staff Astronomer, Astronomy Group
ASTRON, Oude Hoogeveensedijk 4
7991 PD Dwingeloo, The Netherlands
!=============================================================!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listmgr.nrao.edu/pipermail/difx-users/attachments/20160506/192346b4/attachment-0001.html>