OT: Linux CDs

Peter Clinch said:

Richard said:

ITYM FORTRAN ;-) Still the best scientific language around, IMO.

Depends on what science you're doing. If you've got a big array
crunching job where you know how big everything is in advance, want to
let it stew to output bits rather than be interactive or send out
pictures

If I'm doing interactive/picture work, I use IDL :-)

Quoted message said:

and have a good F90 compiler, yes, it's as good as anything
going, /especially/ on parallel processing hardware (but note comment
about "good F90 compiler"😉.

FORTRAN 77 is still Out There being used though, and is ghastly. Sorry,
but COMMON blocks, 6 letter variable names and no dynamic memory
allocation really do make it a Work Of Stan.

Hell yes. Not too long ago I had to take (someone else's) code with
bits of 77 and even 66 and bring it up to speed on an F90 compiler.

R.

Quoted message said:

Next excuses?

Oh give over! Cheap Apple?

eg

http://www.ebuyer.com/customer/products/index.html?action=c2hvd19wcm9kdWN0X292ZXJ2aWV3&product_uid=80436

for £250 inc vat and accessories. And if I wanted a decent spec
machine it wouldn't cost the silly money Apple are asking for
upgrades.

@ The linux tribe, I'm no expert but I did try getting Red Hat on my
laptop only to find out (after a few frustrating days) that 'No, Linux
distros aren't really suitable for laptops' (lots of missing
drivers..) an the same goes for desktop PCs ... finding drivers for
scanners, gfx, mp3 players and so on is generally a pain in the [censored].

Who said that URC was drowned out by a vocal minority who had their
own agenda to push...

Peter Clinch said:

FORTRAN 77 is still Out There being used though, and is ghastly.
Sorry, but COMMON blocks, 6 letter variable names and no dynamic
memory allocation really do make it a Work Of Stan.

Depends on your version of F77 thobut. I am engaged in doing Stuff with VMS
FORTRAN on a daily basis and while some of the ancient legacy code /does/
use COMMON blocks, most of it doesn't. And it allows all sorts of creative
variable names. As to this memory allocation wossname, surely it's the job
of the operating system to do that?

Sadly, when our women and their jobs go to India in the not-too-distant
future I will no longer have a VMS system with which to play.

--

Dave Larrington - http://www.legslarry.beerdrinkers.co.uk/
World Domination?
Just find a world that's into that kind of thing, then chain to the
floor and walk up and down on it in high heels. (Mr. Sunshine)

Dave Larrington said:

Depends on your version of F77 thobut. I am engaged in doing Stuff with VMS
FORTRAN on a daily basis and while some of the ancient legacy code /does/
use COMMON blocks, most of it doesn't. And it allows all sorts of creative
variable names. As to this memory allocation wossname, surely it's the job
of the operating system to do that?

I learned computer programming with VAX FORTRAN and it is indeed miles
better than pure F77, but note that VAX FORTRAN isn't FORTRAN 77, it's a
considerably enhanced superset (with a lot of features that ended up in
F90 IIRC, but it's been a while so ICBW). Similarly, compare Visual
BASIC to BASIC and Turbo Pascal to Pascal.

FORTRAN 77 is a standard (ANSI?) language definition, and a complier
either supports it or it doesn't.

The memory allocation is indeed the OS's job, but how much to ask for
isn't always known in advance of running or (especially) compiling. If
I write a hack to wangle a file, for example, in standard F77 there's no
way I can find out the size of the file on the fly and /then/ allocate
enough, and only enough, memory to deal with it. And since I might be
dealing in files of several hundred Mb with 3D image stacks, that's a
bit of a waste of system resources!

Quoted message said:

Sadly, when our women and their jobs go to India in the not-too-distant
future I will no longer have a VMS system with which to play.

I always got on well with the VAXen I used. Shame the system architect
carried its naturally fragmenting file system onto his subsequent job at
Microsoft though :-(

Pete.
--
Peter Clinch Medical Physics IT Officer
Tel 44 1382 660111 ext. 33637 Univ. of Dundee, Ninewells Hospital
Fax 44 1382 640177 Dundee DD1 9SY Scotland UK
net [email hidden] http://www.dundee.ac.uk/~pjclinch/

Fortran Coloring Book!

Blimey, I'd forgotten the name of that, but it certainly taught me how
to program. "The squaw on the hippopotomus... " - not exactly PC, I
grant you...

I'm up to 15 programming languages at last count, but I reckon my
second-worst ever job was converting a load of F77 written by
non-programmers into C++.

Ian Smith said:

Quoted message said:

Oh, I would expect to use LAPACK/LINPACK for the basic stuff.

Is that on a Mac?

So far as I can see, that approach depends for speed on calling
optimized platform dependant sub-routines rejoicing under the catchy
lable 'the BLAS'. A search for such routines turns them up quite
easily for Compaq, HP, IBM, Intel, SGI, and Sun hardware, but Mac...

Yes, on a Mac and optimised for the vector hardware in G4 and G5
processors. A modern Mac is basically a Unix box with a GUI that
doesn't suck. :-)

See:

<http://developer.apple.com/hardware/ve/vector_libraries.html>
<http://developer.apple.com/hardware/ve/quickstart.html>

Apple provides the libraries as part of the OS.

--
Jose Marques

Richard Bates said:

Right, that's setlled it then. Nobody wants them :-)

Geeky coffee mats they are then ...

Or ...
http://www.komik100.com/komikresim/kresim/bilgisayar/pc1412.jpg

In article <[email hidden]>,

David Martin said:

The only thing I'd want a Windows PC for would be Memory Map, unless similar
software is availabel for Mac/Linux.

Runs fine under VirtualPC ;-)

(yes I know that's cheating)

Shaun

In article <[email hidden]>,

Ian Smith said:

James Annan said:

Oh, I would expect to use LAPACK/LINPACK for the basic stuff.

Is that on a Mac?

LINPACK is often used as a benchmark for testing supercomputers.

The Virginia Tech supercomputer, based on 1100 Macs in a cluster is the
7th fastest supercomputer in the world at running LINPACK.

The same software runs on an iMac.

Shaun

In article <[email hidden]>,

James Annan said:

First couple of googits for blas mac looked promising, but TBH for my
purposes I am not bothered about speed, just getting some robust code
for routine matrix manipulation. It's all downloadable as fortran, the
only point of BLAS is to get it compiled efficiently AIUI.

Bearing in mind it's a lot of vector processing, having it run through a
decent compiler that supports the PowerPC AltiVec instructions would
probably yield both robustness and speed.

eg. using IBM's xlf instead of the gnu tools.

Shaun

In article <[email hidden]>,

Simon Brooke said:

No, we'd all be a lot better off. On security, speed, reliability,
ergonomics and backwards-compatibility, Windows has always been and is
still a very long way behind the state of the art.

To be fair, MacOS only actually won on ergonomics out of those criteria
against Windows before MacOSX came along and even then it was getting a
bit clunky when compared to BeOS.

Mac OS v7.1 was technically as far as they should have pushed it before
replacing the guts underneath that pretty interface. My Amiga at the
time with the same processor used to toast my Mac. And the Mac was a
pain for compatibility with anything.

But anyway, those were the old days and it's irrelevant now. The only
time I run into compatibility problems now is some weirdo PC user that
insists on sending me WordPerfect documents and expects me to be able to
read them.

Shaun