Brian Smith wrote:
> On Tue, 3 Apr 2007, Wayne Boucher wrote:
>
>> We could do with an FAQ-like documentation section which answers these
>> basic questions for beginners. We're going to hopefully (somehow) get
>> the community involved with this (because otherwise we would do nothing
>> but documentation!).
>
> I think this would be a good idea. In this case I think the problem is
> probably that user thinks "I need to pick my 3Ds" but can't find "How to
> pick 3Ds" in the documentation/tutorial and so so doesn't realise that
> they actually want the function that allows them to e.g. "pick and
> assign resonances to their 3Ds based on an initialised 2D".
Hi Brian,
Actually (and this is an aside to the general thread) what my post-doc
wanted to do was pick all the peaks in the 3D spectrum irrespective of
what is in the root 2D spectrum, even if this did generate more picked
peaks than should theoretically be present. This desire might go against
the generally accepted approach; the mind set being that they would
rather over-pick a spectrum than underpick it, I think.
[Actually I am more than a little frustrated that my guys don't put
these questions to the user-group directly. I wonder how many other
potential users are getting discouraged from CCPN-analysis by not
realising or taking advantage of the freely given help that is out there?]
Cheers,
Paul
> New users
> are strongly encouraged to post their questions/frustrations to the list
> rather than be put off. Maybe some of the more experienced among us
> could start building the FAQ on the web site - how's about
>
> http://www.ccpn.ac.uk/ccpn/software/ccpnmr-analysis/faq/how-to-pick-peaks-in-a-3d-spectrum
>
>
> for starters?
>
>
>>> On a related matter, that is inspired in part by the recent
>>> discussion of
>>> OSX implementation performance issues... Is there a recommended
>>> hardware set
>>> for CCPN software on PC/linux platforms? Where are the real bottlenecks?
>>> Is it all in the graphics card?
>
> There's almost too many variables to take into account.
>
> CPU: We get decent performance on all our machines with a Pentium IV
> 2GHz or better CPU. A hyperthreading/dual/dual core CPU can help
> because the python process and the X server can run on separate CPUs and
> so they don't have to wait for each other as much.
>
> RAM: By using the -m switch you can specify a larger contour cache at
> runtime. The larger the cache you have, the less you have to recompute
> contours on the fly and the smoother your experience. 1Gb RAM is
> probably towards the low end of what you would want for regular analysis
> use, more is better.
>
> Graphics card: People are using both ATI and NVidia graphics cards
> running the vendors' hardware accelerated OpenGL with noticeable
> performance enhancements over Mesa (Does anyone have any experience with
> Intel/Matrox etc graphics adapters?). It's probably worth avoiding
> "integrated" graphics cards as they poach RAM from the CPU.
>
--
Paul C. Driscoll
Professor of Structural Biology
---------------------------------------------------------------------
Prof. Paul C. Driscoll |Office (answer)phone: (44)-20 7679 7035
Dept. Biochem. & Mol. Biol. | Department fax: (44)-20 7679 7193
University College London | [log in to unmask]
Gower Street, London WC1E 6BT| http://www.biochem.ucl.ac.uk/~driscoll
|