RE: [ccp4bb]: PC crystallography ---summary

[<Xin.Chen@aventis.com>]

***  For details on how to be removed from this list visit the  ***
***          CCP4 home page http://www.ccp4.ac.uk         ***

Regarding my earlier posting of "portable crystallography", here is a
summary of the replies.  Thanks again to all the respondents!


XIN CHEN
Structural Biology
Aventis Pharmaceuticals
1043 Route 202-206
PO Box 6800, Mailcode N0312A
Bridgewater, NJ 08807
Tel. 908-231-3366
----------------------------------------------
A.G.Evdokimov
P&GP

We have a lot of standard crystallographic software on our laptops. Needless
to
say, Linux, not MS-OS. Apart from a few minor things (such as O8 crashing on
an
attempt to contour a large map cube, which it also does on a desktop)
everything works really well, I have yet to crash CCP4 (previous release) on
the road.

Dell 8100 Inspiron (1.3 GHz Pentium processor, 512 MB DDR RAM) does
reasonably
well in terms of performance. Depending on the task, we see performance
levels
slightly to moderately worse than when we run the same stuff on single
processor machines from our Linux cluster, which uses 1.6 GHz Athlon (1.5 GB
DDR RAM) machines with Asus 266-V boards.

So far, I would not dream of abandoning our lab hardware. Laptop is good for
the road, but it can't replace all the processing power that we need for
efficient work.
-----------------------------------------------
Peter Moody

I have put ccp4, CNS, Shelx and XtalView on two lap-tops (with RH and SuSE 
Linux) without too much trouble. I have not tried using O. My old Toshiba
has an old-type (TFT?) screen, and so models tend to be hard to see when
they are rotating. This made fitteing electron density rather hard. I also
needed a "small screen" version of mosflm to be
practical, but Harry Powell came up with the goods.

My newish  Ergo (Brick in the US I think) has a 15.1" screen and is fine.
It has a 1.2GHz PIII and seems to do all that I ask. Networking is
sometimes a bit of a headache when the interface is built-in. But the
tulip driver sorted it out for me. And the right PCMCIA card should be a
breeze. 

I have processed data in hostel rooms, and refined structures on the
train. I often use the lap-top at home to create figures.

In short it is as practical as PCs in the lab, although you won't find a
dual processor 2.3GHz machine yet! If you really want to use it on the
move, you may find the battery life a problem.

Get loads of memory and as big a disk as you can, I would also make sure
you get a fire-wire interface, incase you want to take fire-wire disks
along to the synchrotron.

I've not tried crystalography with windows, and hope I never have to!

Plugging lap-tops is at a synchrotron is not always trivial though!
----------------------------------------
Marc Saric

Can't speak about solving a complete structure on a PC, but we use our 
Laptop to index and scale our data on-site (i.e. at the synchrotron) and 
it works very well.
Modelling can also be done on a PC, if you don't need dials and probably 
stereo.
Batch-jobs can be run on any fast computer, be it a server or a fast 
desktop.
The difference between a top-notebook and a top-desktop is more or less 
the price. You will pay ~ 100% more for a notebook with comparable 
CPU-power and memory/HD.
Therefore it might be better to use a desktop-PC or server for heavy 
computing and doing only the stuff, which benefits from a portable 
computer, on a Laptop.

In general I don't see any reason why one shouldn't be able to use a 
laptop alone if one has to. Servers in the background are just more 
convenient for long-running tasks and desktop-machines are a lot cheaper.
--------------------------------------------
Klaas Decanniere

You can even have both (on the move + servers) with wireless networks!

I haven't ventured into laptop crystallography for the following reasons
(and I have been tempted):

- battery live and number crunching are a bad combination in every possible
way

- backups are an important issue (laptops get dropped - hard drives don't
like that - you don't want to loose your high impact structure because you
had to run to catch that train)

-  no high-end openGL graphics card, fiddly
mouse/trackball/pointing thingy, so running O or TurboFrodo is
impractical

- no way to get a 3 wavelength finesliced MAD dataset from a synchrotron
trip on your laptop hard drive for processing

- data processing software likes big screens

- fast SCSI is what it sounds like: fast. An important issue for large data
files, and not practical on laptops

- a high-end server plus a medium-end laptop is probably cheaper than
a high-end laptop

- (allmost) all laptops are prorietary - upgrades (memory!) and repairs will
be
expensive


So I stick to word processing, preparation of figures, telnet, netscape  and
giving
presentations while I'm on the move

Run the job on your server in the lab, pick up the results when you have the
chance via ftp, look at them and prepare the next step, and submit the new
job via a telnet session.
------------------------------------------
Gerry McDermott

I often use a Dell 1.2 Ghz laptop, or an Apple Titanium to do
crystallography on the move.  Both machines run O or Pymol graphics fast,
and can do ''calculations' as fast as servers from only a few years
ago......the only downsides.  Its difficult to do any manipulations on the
graphics programs without using an additional three-button mouse.  And, the
Dell, although fast, is heavy and the batteries short lived once you crank
through some calculations.

Otherwise, for the most part, its not particularly slower to work on a well
spec'd out laptop (1 Gb Ram anda decent video card are important).
Especially, if you are at a synchrotron - where you have access to all of
your scripts and files, and can take all of your data away on a firewire
drive.  For a total outlay of a coupe of thousand, this is so much better
than using tapes, restoring the data when you get home, etc. etc. etc.
----------------------------------------------
Harry Powell

I'd say it was dead easy and perfectly practical on a laptop to do your
crystallography anywhere you choose - but you'd probably still want a
darkened room for model building... I regularly use a clunky 'old' (2
years) Powerbook for integrating/scaling/etc "on the road"; I don't often
get the chance to do the building etc these days!
-----------------------------------------------
Flip Hoedemaeker

CCP4i works fine on my 2 year old win2000 laptop (Sony VAIO SR1K, Intel PIII
500MHz, 128kB), albeit a little slow compared the latest processors.