[Mrtrix-discussion] mrtrix on xeon phi?

[Donald Tournier]

Hi Luis,

We had a brief look at this following your email. Sounds essentially like
this PCI card contains a full Linux system housed within your own computer,
which can be accessed and managed via TCP/IP. If that's the case, I don't
anticipate there would be any issues running MRtrix on it - you'd just need
to set up the NumberOfThreads entry in its config file to something
large... This is assuming you can simply log onto the card just as if it
were another computer on your network. However, if the only way to manage
the software is via MPI, then there'd be no way to run MRtrix without some
serious changes to the code...

As to whether it's worth the investment, I'm not sure. There's only 8GB of
RAM on the card, which may turn out to be a limiting factor down the track.
The number of threads is awesome, but the speedup is less impressive than
might have been anticipated (2.5x compared to a dual-socket 8-core Xeon
setup). The other issue with that many threads is the possibility of mutex
collisions (too many threads trying to hold write lock at the same time),
which might impact performance - although this can probably be optimised
relatively easily. Probably still worthwhile despite all this though, since
this card costs about the same as a single 8-core Xeon...

This is just from a brief look at it, so don't take my word for it - I
could have this all backwards...

Cheers,

Donald.


On 27 February 2013 04:43, Luis Concha <lconcha at unam.mx> wrote:

> Dear mrtrix users:
>
> The Intel Xeon ph<http://www.intel.com/content/www/us/en/processors/xeon/xeon-phi-detail.html>i
> seems like a sweet ride. With 60 cores and 240 threads, it is tempting to
> run tractography at a massive scale. Would this be straightforward to do
> with the current mrtrix code?
>
> In general, do you think this is a good investment? I would imagine other
> medical imaging software packages would also benefit from so many cores.
>
>
> Dr. Luis Concha
> Instituto de Neurobiología
> Laboratorio C-12
> UNAM, Campus Juriquilla
> Boulervard Juriquilla 3001
> Juriquilla, Querétaro.
> C.P. 76230
> México
> Tel (442) 2 38 10 53
> Fax (442) 2 38 10 46
> www.inb.unam.mx
>
> _______________________________________________
> Mrtrix-discussion mailing list
> Mrtrix-discussion at www.nitrc.org
> http://www.nitrc.org/mailman/listinfo/mrtrix-discussion
>
>


-- 
*Dr Jacques-Donald Tournier
*
Research Fellow

The Florey Institute of Neuroscience and Mental Health
Melbourne Brain Centre - Austin Campus
245 Burgundy Street
Heidelberg  Vic  3084
Ph:  +61 3 9035 7033
Fax:  +61 3 9035 7307
www.florey.edu.au
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.nitrc.org/pipermail/mrtrix-discussion/attachments/20130304/399e3eae/attachment.html