Sunday, 11 September 2011

Windows 7 and Xtal Software

From: Jacob Keller
Date: 29 August 2011 03:23
Dear Crystallographers,

are there any additional problems or known issues running ccp4 or
other xtal software on windows 7 (beyond those of Vista, etc.?) Your
input would be really appreciate before I sink my own personal $$$
into a new laptop....

Jacob Keller



--
*******************************************
Jacob Pearson Keller
Northwestern University
Medical Scientist Training Program
*******************************************

----------
From: Nat Echols
Phenix, ARP/wARP, and HKL2000 do not run on Windows.  I'm pretty sure none of Global Phasing's software does either (aside from web interfaces).


-Nat

----------
From: Ian Tickle
Correct, and you can add XDS to the list.  Also a number of data processing utilities such as adxv & xds-viewer are only available for Linux/OSX (though iMosflm which is on Windows could be used instead).

Cheers

-- Ian

----------
From: Roger Rowlett
The CCP4 suite runs very well in a virtual machine (e.g. VirtualBox) in Windows. You can share folders between VBox and Windows so you can run graphically intensive apps like Coot and Pymol natively under Windows but still use data from your Linux VM.
Roger Rowlett

----------
From: George M. Sheldrick
The current SHELX binaries (including the beta-test multi-CPU SHELXD) all
appear to run fine under Windows 7. There is no need to use a virtual box etc.
George
--
Prof. George M. Sheldrick FRS
Dept. Structural Chemistry,
University of Goettingen,
Tammannstr. 4,
D37077 Goettingen, Germany

----------
From: George M. Sheldrick
It is simply a result of the 'zero dependency' philosophy. In other words, the
exact opposite of current trends in crystallographic computing (e.g. Phenix/CCTBX).

George

On Mon, Aug 29, 2011 at 12:39:16PM -0500, Jacob Keller wrote:
> You know, why does your software always seem so clean? Was it
> something about the punch cards?
>
> Jacob

----------
From: Paul Smith <paul@paladinscientific.com>
Date: 30 August 2011 00:01
To: CCP4BB@jiscmail.ac.uk

Okay then,

Here's the plan for all the software developers out there:

1) backport cctbx to fortran (preferably F77) and include all subroutines inline to avoid the use of any external libraries whatsoever
2) ditch all gui support or, from scratch, develop a gui front-end that uses none of the following: Qt, Ruby, Perl, Python, TK/TCl, etc.  This gui must compile and run on all mainstream hardware on all major operating systems.  The custom gui might also need a custom driver for maximizing the capabilities of modern GPU's for 3D work, but shouldn't make use of any existing shading/tiling/rendering methods (like openGL).
3) scratch all binary formats (mtz,ccp4map,etc.) due to interoperability issues/dependencies.  Everything in flat text (if you like, all variables can have four letters and can be followed by a flag/switch consisting of an integer or two, perhaps negative, to control software behavior).
4) abandon rapid software development afforded by modular, object oriented programming.

Sounds good to me.

Seriously however, I do like how well-coded monolithic executables simply work once compiled without fuss.  I also like the speed and power afforded by using a well thought out toolkit of practical modules, a la PHENIX.  I guess I can't have it all.  Personally, if you really need windows, I second the idea of a VM running a linux environment.  It's vastly simpler to install mature linux binaries within a VM then fight to get all of modern crystallographic software to run under windows.  Even better, the other way around -- run linux native and windows in a VM.

For the record:
Shelx is awesome
Fortran is a perfectly good programming language
I keep a slide rule in my desk.

--Paul

Paul Smith, Ph.D.
- Memorial Sloan-Kettering Cancer Center
- New York Institute of Technology
- www.paladinscientific.com



----------
From: William Kennedy
the question, then, is how fast is your FFT on a slide rule?

Dexter Kennedy, MD
Thumbed from my iPhone

----------
From: Nian Huang
A dual boot laptop is all you need. I always reinstall the windows to get rid of bloatware anyway. If you are going to buy a mac, you can also try the triple boot, but I don't think anybody is doing it. Although it is very convenient, a virtual machine will affect the performance of the software. Nowadays booting machine is very fast using a good SSD (under 7 second). So it is really not a big trouble comparing before. I am using a sub $400 laptop, and everything runs really well under Ubuntu including model building software.

Nian





----------
From: Antony Oliver 


Erm, somewhat confused — if you are going to buy a Mac — why would you need (or want!)  a triple boot system? It all seems to work just fine on OS X.

Tony.

----------
From: Nian Huang
There are some windows only software, for example 3ds Max. But you have a point, you probably don't need linux system in this case.

Nian


----------
From: Pete Meyer
I know you were joking, but I'm all in favor of dropping gui's for tasks that don't involve dealing with graphical data (models, maps or diffraction images).  It's usually faster to learn new software with a gui; but it often seems to result in users not reading the documentation - which may contribute to newcomers to crystallography using software to produce structures without understanding what the magical black box is doing (aka the clickey-button effect; which seems to pop up every time there's a discussion of a retracted paper/problematic structure).

But I don't have a slide rule, so my opinion on these matters may be invalid.

Pete


----------
From: Ed Pozharski
second that.  I was about to say "while it is not expected that everyone
practicing crystallography should master the use of command line", but
the question is why not?  It's really not that hard, makes a lot of
tasks easier and as Pete correctly points out, forces users to RTDM.
But the "tablet mentality" will, of course, win.

As for my recent addiction to the SSRL's autoxds script - guilty as
charged. :)

--
Oh, suddenly throwing a giraffe into a volcano to make water is crazy?
                                               Julian, King of Lemurs

----------
From: Roger Rowlett
Personally, I like having the GUI front end for training and education, especially for undergraduates. It has made protein XRD much more accessible as a tool for many labs that would otherwise find the barrier for entry very high. In the "old days,"--8 years ago (ha ha)--scripting from the command line was pretty much the only way to run most protein XRD software. I think scripting is much more powerful, and allows for a nicely pipelined and controlled refinement workflow, but it is easier to train undergraduate students, and get them productive quickly, with the GUI. The emergence of Coot and the CCP4i front-end was a godsend for the undergraduate research laboratory. The GUI is kind of like a checklist for protein XRD tasks. But just because you have a GUI doesn't mean you should ignore the log files...sometimes my new students think that is optional. ;)

Having said all that, the thought of running protein XRD software in Windows (except when that is the only option, e.g. CrysalisPro) is not a cheery one: Windows scripting capability is clumsy, and boy, is it ever slow! At one point, I think I timed a typical Phaser or EPMR job we ran at 4X slower in WinXP than in Linux. It was the difference between about an afternoon and overnight. Computers are faster now... :)

Cheers,

_______________________________________
Roger S. Rowlett
Gordon & Dorothy Kline Professor
Department of Chemistry
Colgate University
13 Oak Drive
Hamilton, NY 13346

----------
From: Jacob Keller
Dear Crystallographers,

once again I am filled with graditude to this list--there were many
helpful responses and even some geek humor. I have decided to go with
a dual-boot windows7/linux, which seems easy enough. All the best, and
thanks everyone for your quick and helpful advice.

Jacob Keller

No comments:

Post a Comment