My advice:
Embrace virtualization for all tasks except interactive 3D visualization.
If you're not yet familiar with VMware, Parallels, or open-source work-alikes, then it is high time you joined the revolution --- the rest of us have been running Mac OS X, Linux, and Windows on the same hardware *simultaneously* for years now.
So long as you can dedicate at least 1-2 GB of RAM per running OS, it works great,
and you get many other benefits from breaking the link between OS and hardware
(e.g. easy backup & restore, better security, snapshots, test before you
upgrade, trivial migration to new hardware, never reinstall, never have to buy new
software, etc).
Unfortunately, however, for interactive 3D visualization, you must still choose
the host OS platform which runs your favorite visualization tools best, since
effective virtualization of OpenGL remains an elusive goal.
Personally, I prefer Mac hardware and typically use Mac OS X as the host operating
system. But, as I write this, I am also simultaneously running Windows for Excel
& PowerPoint and GNU/Linux for open-source software development.
Of course, there are good reasons for running other host OS platform instead, such as to obtain native 3D graphics support under Linux or ActiveX Controls under Windows.
The point is, you don't have to choose just one OS anymore. This survey seems to suffer from an outdated assumption.
Cheers,
Warren
________________________________________
From: CCP4 bulletin board [mailto:ccp...@jiscmail.ac.uk] On Behalf Of Roger
Rowlett
Sent: Friday, May 01, 2009 8:04 AM
To: CCP4BB@JISCMAIL.AC.UK
Subject: Re: [ccp4bb] Computer hardware and OS "survey"
Well, Coot, O, Pymol, CNS, and CCP4i, as well as Open-EPMR all have Windows
versions. The main issues with a Windows workflow are (1) jobs will run
significantly slower than in Linux, and (2) the DOS command shell is not as
powerful as Linux, although it can be extended by installing DOS versions of
Linux commands and utilities. You will also lose access to a number of
Linux(Unix)-only XRD tools, but those are getting fewer each year. It's also
easier, more stable, and more secure to set up a laboratory data server in
Linux than in Windows. You will also find that you can get excellent computing
performance out of fairly modest hardware in Linux compared to Windows.
I'm not sure there is much "institutional support" required for Linux if you
know how to install your own OS and software. All I need from my networking people is a
hole in the firewall for my MAC address and SSH port. After that, there is not much for
IT to do for me other than stay out of the way. Ubuntu has made it a lot easier than it
has been to maintain your own Linux systems, but I'm still currently wedded to Fedora.
The main Linux headache is hardware support, especially printers and graphics drivers for
Nvidia cards, but even that is relatively painless now.
Cheers,
________________________________________
Roger S. Rowlett
Professor
Colgate University Presidential Scholar
Department of Chemistry
Colgate University
13 Oak Drive
Hamilton, NY 13346
tel: (315)-228-7245
ofc: (315)-228-7395
fax: (315)-228-7935
email: rrowl...@mail.colgate.edu
Link,Todd M wrote:
My home institution, in effort to cut costs, is making an effort to push those of us on Macs onto PCs. Up till now they have been very generous via a lease program for computer hardware, but that is changing given the current economics. The institution currently does not support Linux so we are limited to Mac and Windows OS.
We certainly make use of William Scotts crystallography on OS X (thanks so much!) so
our main argument is that we would have far more support "out there" for
crystallography on the Mac than we would have for on Windows. But to be fair (and
hopefully bolster our argument) I should find out if that is true. I did not find an
equal web support page for Windows.
A volunteer survey will be distorted (probably by Mac fanboys like me) so I am
asking for peoples best guesstimate as to what % use of Mac, Windows, or Linux
is out there for data processing and model building. Our core programs are
coot, o, pymol, cns, and ccp4 but we certainly make occasional use of other
crystallography programs out there (solve, epmr...)
Also what are the relative crystallography support for Mac vs. Windows.
Thanks in advance.
Todd
