Each experiment has 6 pigeons and panels, and we run these simultaneously.
I’ll take the example of the machine that runs Experiments 2, 4, 5 and
10 (one more to be added). The I/O for this set consists of 15 DIG-716B cards
set to the appropriate memory locations. Each of the 716Bs (which have 8
inputs and 16 outputs) controls two pigeon panels, so each one has 4 inputs
and 8 outputs (we wired the connections ourselves, but I am sure that MED
Associates could easily provide appropriate connection leads). Thus, the
first 3 716s run the 6 birds in Experiment 2. The 5 experiments are run s
uccessively – Pigeons 21-26, then Pigeons 41-46, and so on. We use
MED-PC 4, build 46.
Because you can’t (apparently) set up 36 boxes in MED-PC configuration
(I’d be happy to be shown to be wrong!), we have to use special
procedures to load the appropriate configuration (.dta) files for each
successive experiment, using the /INSTALL= command line addition when
loading MED-PC. There is a bug in the interpretation of this in Build
46 (and probably in previous builds) that fails properly to interpret
commands of the form
/INSTALL=”C:\MED-PC IV\filename.dta”
and this command will only work in this way
/INSTALL=filename.dta
and for this to work, you have to have moved to the MED-PC IV directory
before starting MED-PC.
This fragment is how we do this for Experiment 2 in a batch file:
cd c:\med-pc IV
start/wait MEDPC_IV.exe /I /INSTALL=MPC2INST2.dta
the cd command moves the focus to the MED-PC IV directory so that
MPC2INST2.dta can be found without drive and directory information.
Start/wait starts MED-PC and waits until the software has completed
running before moving on to the next line in Batch (without this, batch
files will do all the commands on all lines without waiting). Thus, the
exit on completion in MED-PC has to be set up for all MPC running programs.
/I forces MED-PC to use the interface – sometimes, we have found,
it can fail to find it when it is there.
The Batch file that runs all of the experiments (Starter.bat) is:
del c:\birds\data\*.0
cd c:\med-pc IV
start/wait MEDPC_IV.exe "c:\MED-PC IV\macro\macEXPT2.mac" /I /INSTALL=MPC2INST2.dta
start/wait MEDPC_IV.exe "c:\MED-PC IV\macro\macEXPT4.mac" /I /INSTALL=MPC2INST4.dta
start/wait MEDPC_IV.exe "c:\MED-PC IV\macro\macEXPT5.mac" /I /INSTALL=MPC2INST5.dta
start/wait MEDPC_IV.exe "c:\MED-PC IV\macro\macEXPT10.mac" /I /INSTALL=MPC2INST10.dta
start/wait c:\birds\expt5\b5anal.exe
start/wait c:\birds\expt5\Crender5.exe
start/wait c:\birds\expt2\cb2anal.exe
start/wait c:\birds\expt2\Cmove2.exe
start/wait c:\birds\expt4\cb4anal.exe
start/wait c:\birds\expt4\Crender4.exe
start/wait c:\birds\expt10\cb10anal.exe
start/wait c:\birds\expt10\convert.exe
(without the carriage returns on the starting MED-PC lines)
In this way, the (current) 4 experiments are run successively, and
then a series of PowerBASIC executables are run that do things like
simple analyses, and moving, backing up, and truncating data files
made by MED-PC( because we use large arrays that are never completely
filled).
For exch experiment, an appropriate macro is set up – for example,
for Experiment 2, we have:
LOAD BOX 1 SUBJ 1 EXPT 2 GROUP 0 PROGRAM EXPT2
FILENAME BOX 1 BIRD21.0
LOAD BOX 2 SUBJ 4 EXPT 2 GROUP 0 PROGRAM EXPT2
FILENAME BOX 2 BIRD24.0
LOAD BOX 3 SUBJ 2 EXPT 2 GROUP 0 PROGRAM EXPT2
FILENAME BOX 3 BIRD22.0
LOAD BOX 4 SUBJ 5 EXPT 2 GROUP 0 PROGRAM EXPT2
FILENAME BOX 4 BIRD25.0
LOAD BOX 5 SUBJ 3 EXPT 2 GROUP 0 PROGRAM EXPT2
FILENAME BOX 5 BIRD23.0
LOAD BOX 6 SUBJ 6 EXPT 2 GROUP 0 PROGRAM EXPT2
FILENAME BOX 6 BIRD26.0
EXIT_WHEN_DONE ON
START BOXES 1 2 3 4 5 6
(again without the carriage returns). Similar macros are written
for the other experiments.
The odd order is because the first 716B runs Pigeons 21 and 24,
just because of the physical way we have the cages in the lab –
it’s easier to run the partial DB 25 connection between 21 and 24
than between 21 and 22.
An example of a .dta file may be useful – this one Experiment 4,
which is run from the 3rd to the 5th 716Bs:
MPC INSTALLATION 2.0
RESOLUTION 10
NUMBOXES 6
TIMERADD 768
LONG_NAME 0
ENDTIME 1
DRIVE C:\birds\data\
FLOPPY 0
BACKUPS 0
AUTODUMP 0
NAMEFORMAT C
PRINTTYPE 4
NAMECHAR !
BKUPFREQ 10
IRQ 3
RMAX 1 4
RMAX 2 4
RMAX 3 4
RMAX 4 4
RMAX 5 4
RMAX 6 4
INPUT 1 1 783 0 0 1 1
INPUT 1 2 783 0 0 2 1
INPUT 1 3 783 0 0 4 1
INPUT 1 4 783 0 0 8 1
INPUT 2 1 783 0 0 16 1
INPUT 2 2 783 0 0 32 1
INPUT 2 3 783 0 0 64 1
INPUT 2 4 783 0 0 128 1
INPUT 3 1 784 0 0 1 1
INPUT 3 2 784 0 0 2 1
INPUT 3 3 784 0 0 4 1
INPUT 3 4 784 0 0 8 1
INPUT 4 1 784 0 0 16 1
INPUT 4 2 784 0 0 32 1
INPUT 4 3 784 0 0 64 1
INPUT 4 4 784 0 0 128 1
INPUT 5 1 785 0 0 1 1
INPUT 5 2 785 0 0 2 1
INPUT 5 3 785 0 0 4 1
INPUT 5 4 785 0 0 8 1
INPUT 6 1 785 0 0 16 1
INPUT 6 2 785 0 0 32 1
INPUT 6 3 785 0 0 64 1
INPUT 6 4 785 0 0 128 1
MAXOUT 1 8
OUTPUT 1 1 792 1 6 1 1
OUTPUT 1 2 792 1 6 2 1
OUTPUT 1 3 792 1 6 4 1
OUTPUT 1 4 792 1 6 8 1
OUTPUT 1 5 792 1 6 16 1
OUTPUT 1 6 792 1 6 32 1
OUTPUT 1 7 792 1 6 64 1
OUTPUT 1 8 792 1 6 128 1
MAXOUT 2 8
OUTPUT 2 1 792 1 7 1 1
OUTPUT 2 2 792 1 7 2 1
OUTPUT 2 3 792 1 7 4 1
OUTPUT 2 4 792 1 7 8 1
OUTPUT 2 5 792 1 7 16 1
OUTPUT 2 6 792 1 7 32 1
OUTPUT 2 7 792 1 7 64 1
OUTPUT 2 8 792 1 7 128 1
MAXOUT 3 8
OUTPUT 3 1 792 1 8 1 1
OUTPUT 3 2 792 1 8 2 1
OUTPUT 3 3 792 1 8 4 1
OUTPUT 3 4 792 1 8 8 1
OUTPUT 3 5 792 1 8 16 1
OUTPUT 3 6 792 1 8 32 1
OUTPUT 3 7 792 1 8 64 1
OUTPUT 3 8 792 1 8 128 1
MAXOUT 4 8
OUTPUT 4 1 792 1 9 1 1
OUTPUT 4 2 792 1 9 2 1
OUTPUT 4 3 792 1 9 4 1
OUTPUT 4 4 792 1 9 8 1
OUTPUT 4 5 792 1 9 16 1
OUTPUT 4 6 792 1 9 32 1
OUTPUT 4 7 792 1 9 64 1
OUTPUT 4 8 792 1 9 128 1
MAXOUT 5 8
OUTPUT 5 1 792 1 10 1 1
OUTPUT 5 2 792 1 10 2 1
OUTPUT 5 3 792 1 10 4 1
OUTPUT 5 4 792 1 10 8 1
OUTPUT 5 5 792 1 10 16 1
OUTPUT 5 6 792 1 10 32 1
OUTPUT 5 7 792 1 10 64 1
OUTPUT 5 8 792 1 10 128 1
MAXOUT 6 8
OUTPUT 6 1 792 1 11 1 1
OUTPUT 6 2 792 1 11 2 1
OUTPUT 6 3 792 1 11 4 1
OUTPUT 6 4 792 1 11 8 1
OUTPUT 6 5 792 1 11 16 1
OUTPUT 6 6 792 1 11 32 1
OUTPUT 6 7 792 1 11 64 1
OUTPUT 6 8 792 1 11 128 1
INPORT 0 783 0 1
INPORT 0 784 0 1
INPORT 0 785 0 1
OUTPORT 1 792 6 1
OUTPORT 1 792 7 1
OUTPORT 1 792 8 1
OUTPORT 1 792 9 1
OUTPORT 1 792 10 1
OUTPORT 1 792 11 1
TESTPORT 783 -1 1
TESTPORT 0 0 2
TESTPORT 0 0 3
TESTPORT 0 0 4
INPORTMIN 783
INPORTMAX 785
OUTPORTMIN 792
OUTPORTMAX 792
INRACKMIN 1
OUTRACKMIN 1
Notice the input memory locations starting at 783 (the first 716
is at 780), and the memory locations and offsets for the outputs.
So, how do we get the STARTER.BAT above to start at the right
time? We use the MS task scheduler (we use Win2000) to run this
file at an appropriate time. This is not ideal, but seems to be
all that is available (other schedulers, like AT, have the same
problems below). The problem seems to be this: Our machines are
connected to our network, and we have to have automatic updates
from MS turned on. Even though we set this to “advise us when
new updates can be downloaded”, even this process seems sometimes
to kill the task scheduler – requiring the password again to be
supplied before it will start the batch file again. This doesn’t
happen when we are not connected to the network. Thus, about
once every month or two, the experiments are not started and we
lose a set of data. But this seems a small price to pay – the
benefits of this system of running are (1), it’s accurate every
time; (2), no one has to mess around actually running the pigeons;
(3), the pigeons get handled less (just weighed once a day);
(4), the data are available to us when we come to work in the
morning; and (5), with the machines connected to the network,
we use another task scheduler set up to back up all our data
and running files each day to the network (using Xcopy). The
drawback is the cost of all the MED interfacing, but we try
to use this efficiently.
If any reader can see better ways of doing this, please
post to this website.
1 With thanks to Tom Tatham who has given us critical help
on a number of occasions, and Doug Elliffe who is the co-developer
of the above procedures.
