atlc - Arbitrary Transmission Line Calculator.

Program for analysing arbitrary transmission lines and directional
couplers. See HTML documentation in docs/html-docs

*********************************************************************
IMPORTANT - if altc won't compile on your system, configure with the
option --without-hardware-info. It should then compile.
*********************************************************************

atlc is available as source code in a file atlc-version.tar.gz. Some
older binary releases are in existance but I would advise against
using version of atlc prior to 4.6.0 for multiple dielectrics.

See the file INSTALL if building from sources. Online documentation may
be found at http://atlc.sourceforge.net/

This is the directory layout. .
./m4                     // Some autoconf macros
./docs                   // Various bits of documentation.
./docs/html-docs         // As on the atlc web site at atlc.sourgeforge.net
./docs/html-docs/jpgs    // JPGS needed for the html docs above.
./docs/qex-december-1996 // A old paper I wrote for QEX, gives the theory.
./docs/theory            // Some very rough ideas I have (Word Format)
./man                    // Standard UNIX man pages.
./examples               // Some examples
./tests                  // Self-tests and a benchmark.
./tools                  // Some tools that are useful for checking the code
./tools/src              // Source code for above
./src                    // The main source code
./tmp                    // Used during tests to dump data.

After building (see INSTALL, which I've modified to be specific to
atlc) you should run 'make check' after running 'make'.
One built, if you want to install the examples and binaries, run
'make install'. As the examples are large (~220 Mb) you might
choose not to install them. In which case run 'make install-exec'
rather than 'make install'. If you later want to install the
examples, run 'make install-data'

If wishing to contact me by email, mail me on drkirkby at gmail.com

Dr. David Kirkby, G8WRB. 27th March 2011 Re: atlc-4.6.1

Unless atlc is configured with the --disable-hardware-info option
(only do this if it fails to compile properly), a small benchmark at the end
tries to gather information about the hardware. The following shows
what information is obtained on each system.

If anyone can add to the information gathered on a system, or add
information on a new system, please email me and discuss how to
do this. Basically a C file in the tests subdirectory needs expanding
or adding (say try_foo.c), where the system tries to gather data on
system foo.

Feature                                 Operating system

                           AIX HP-UX IRIX Linux NetBSD OpenBSD Solaris Tru64
Harware provider            Y                                     Y
Hardware platform                                                 Y      Y
Machine                     Y    Y          Y                     Y      Y
Sysname                     Y    Y          Y                     Y      Y
Release                     #1   Y          Y                     Y      Y
Nodename                    Y    Y          Y                     Y      Y
#CPUs supported                  Y                                Y      Y
#CPUs online                Y    Y                                Y      Y
CPU type                         #2         Y                     Y      Y
FPU type                         #3                               Y
Speed of CPU(s)                  Y                                Y      Y
RAM size (Mb)               Y    Y          #4                    Y      Y
L1 data cache size
L1 instruction cache size
L2 cache size

Feature                                 Operating system

                           UNICOS
Harware provider            Y
Hardware platform
Machine                     Y
Sysname
Release
Nodename
#CPUs supported
#CPUs online
CPU type
FPU type
Speed of CPU(s)
RAM size (Mb)
L1 data cache size
L1 instruction cache size
L2 cache size

Notes:

#1 On AIX, the release is reported as 2, not 5.2 as expected.
#2 on HP-UX, the CPU is reported as  a number, which might not be that
useful.
#3 on HP-UX, the FPU is reported as  a number, which might not be that
useful.
#4 On a machine with 128 Mb of RAM, the ram is reported on Linux as
124 Mb, not 128 Mb as expected.

Want to help with atlc? Here's how.

1) First of all, please test atlc on your platform. Let me know
if it passes or fails and if possible give me the output
of a log file
configure > log
make >> log
make check >> log

and send me the log

2) If anyone can write any code to obtain information about
hardware on any systems, it would be much appreciated.
The format should be very clear form the file
tests/benchmark.c. This looks for a number of files
try_linux.c, try_solaris.c etc. These files are expected
to fill in the structure computer_data


struct computer_data{
char mhz[MAX_SIZE];                  /* Speed in MHz. Note its a string */
char cpus[MAX_SIZE];                 /* Number of CPUs running          */
char max_cpus[MAX_SIZE];             /* Maximum number of CPUs support by system */
 char sysname[MAX_SIZE];              /* System name - got by uname */
 char nodename[MAX_SIZE];             /* nodename - got via uname            */
 char release[MAX_SIZE];              /* Got via uname, but often wrong      */
 char version[MAX_SIZE];              /* Version of the OS. Sometimes wrong. */
 char machine[MAX_SIZE];              /* Machine name                        */
 char cpu_type[MAX_SIZE*3];           /* The CPU type                        */
 char fpu_type[MAX_SIZE*3];           /* The FPU type                        */
 char memory[MAX_SIZE];               /* The memory in Mb                    */
 char hw_provider[MAX_SIZE];          /* The hardware provider - Sun, IBM etc */
 char hw_platform[MAX_SIZE];          /* The hardware platform - eg sun4u    */
 char t1[MAX_SIZE];                   /* Time in s to run benchmark single threaded */
 char t2[MAX_SIZE];                   /* Time in s to run benchmark multi threade
 d */
 char speedup[MAX_SIZE];              /* t1/t2 */
 char eff[MAX_SIZE];                  /* The efficiency - t1/(t2*cpus)     */
 char L1data[MAX_SIZE];               /* Size of level 1 data cache in kb  */
 char L1instruction[MAX_SIZE];        /* Size of level 1 instuction cache in kb *
 /
 char L2[MAX_SIZE];                   /* Size of L2 cache                  */
 }a;

with missing information. So if you know how to find the speed of
a linux box, please change try_linux.c so it has the speed in the
structure element 'data.mhz' All information must be stored as
one long string. So if fpu type should be a 'Athlon XP2800', this
then data.fpu_type should say 'Athlon_XP2800', with the underscore
added.

This really needs doing for Linux and the BSD versions. IRIX I
can do myself at some time.

3) Add a GUI

4) Provide an HP compiler for HP-UX.

5)

Dr. David Kirkby, 15th August 2003.

ATLC should run fine on any linux system, though
no effort has been taken to gather information
about linux systems. (For some operating systems
ATLC reports the CPU type, cache sizes etc).

'atlc' *did* have C code to allow it to use the Message Passing Interface (MPI),
which whould allow 'atltc' to run on distributed processors on a network
of computers. In order to do this, you needed to install the MPICH
library from
http://www-unix.mcs.anl.gov/mpi/mpich/indexold.html

However, at the time of writing, This has been disabled and so there
is not a lot of point in configuring atlc for MPI support. MPI support
is a number one pririty now, so soon a better implementation will be
added to atlc.

Dr. David Kirkby, 15th Auguest 2003, Re: Version 4.3.7

1) Sun's make programme in /usr/ccs/bin/make is not able to compile this
properly, giving an error message like

mksh: Fatal error in reader: = missing from replacement macro reference

Another problem found on the make in Solaris 2.5 is that it don't know
how to make acx_mpi.m4. That file is included in the distribution for
future use, but there is no reason for the program to do anything with
it.

Use GNU make to build atlc. This is often installed on systems as gmake.
Other make versions like smake may work, but I've never tried them.

3) If you have a copy of Sun's cc compiler, the following
CFLAGS are suggested. Using Sun's compiler will give about a 25% reduction
in run times.

'-dalign -fns -fsimple=1 -xalias_level=basic -xdepend -xlibmil -xO5 -I/usr/local/include -R/usr/local/lib'

If you find a better set of options for Sun's cc, that still allow all
the tests to pass, then please let me know.

Dr. David Kirkby, 16th October 2003 Re: atlc-4.5.1

Whenever possible support will be provided free of charge as soon as
possible for any amaateur or commerical use of this software. Just drop
me an email at drkirkby@ntlworld.com and/or davek@medphys.ucl.ac.uk

For extended technical support in using the program in a commerical
enviroment, I would expect some payment.

Dr. David Kirkby, 15/12/02.

This file is README.threads

Prior to verison 4.5.1, atlc could be configured to use
mutltiple-threads to exploit multi-processors. This has
been disabled, as I have some concerns the code may not
be 100% correct and whilst is it unlikely, the results
could be wrong. It is very unlikely, but its a chance
I don't want to take, so multi-threaded support has
been disabled. A new algorithm has been worked out -
it just needs to be coded. Howfully this will be done
by early Feburary 2004 at the latest.

Dr. David Kirkby, 7th December 2004. Re: atlc-4.6.0

atlc is primarily a UNIX program. Some Microsoft Windows
binaries have been produced in the past, one of which
was atlc-4.4.4_Win32.zip. However, some problems were
found in atlc with multiple dielectrics. As such I've
removed the windows binaries.

If someone wants to build some, based on the latest
code I'll put them up again, but whilst the code
is not working optimally with multi-dielectrics, it
is not seriously broken as it was before.

The best solution for Windows uses is to use
Cygin, which gives you a Linux environment on a
Windows computer. ATLC should build fine on that,
though I have never tried it.


Dr. David Kirkby, G8WRB,  27th March 2011
Re: atlc-4.6.1