perm filename COOK.MAI[1,VDS] blob sn#282955 filedate 1977-05-17 generic text, type C, neo UTF8
COMMENT āŠ—   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002		Thanks for  prompting me  with  your call.   I will  call  Suzanne
C00006 ENDMK
CāŠ—;
	Thanks for  prompting me  with  your call.   I will  call  Suzanne
tomorrow and wish her well.  I have rounded up all the required  documents
and they will be  on their way tomorrow.   The symbol between PRELOAD  and
WITH is underbar.   You asked  about what system  I intended  to use  this
program with.  It must run  on an LSI-11 in  stand alone mode.  Thus,  the
entire calibration  routine  should  use  only  an  LSI-11,  the  extended
instruction chip  (incl.  floating  point instructions),  and  8k  of  RAM
memory, although a  program which runs  in only 4k  would be even  better.
You will be allowed to use a standard TI terminal, and my interface to the
force balance.  Thats all.   No RT-11 or other  operating system would  be
available, as I  hope to  have this  program operate  in the  field, by  a
customer.

	I talked  with Bruce  and  we discussed  the  use of  Fortran  for
Decompose and Solve.   Bruce thinks  that the  only 11  Fortran around  is
intended to operate with  RT-11 or similar  disk based operating  systems.
Thus the assembled code would probably make many system calls.  This would
be difficult to fix for use in my simple system.  He thinks that you would
eat up lots of time getting things to work that way, although it would  be
faster to code for a  person with an RT-11 system.   Id like to make  this
code run on the system described above, which is the same as the arm uses.

	Bruce estimates that  it should take  about 100 hours  to get  the
whole program  running in  the LSI-11,  coding everything  in PALX.   With
Fortran you could same  a lot of  time, but it wouldnt  run in the  simple
system.  Bruce also says that if you want, he once worked on conversion of
floating point numbers to output strings  and visa-versa.  The code is  un
debugged, but he will round it up  if you need it.  This may speed  things
up a  bit, but  you will  have  to debug  it.  Bruce  has also  gone  over
WRIST.SAI, and  has  commented a  copy.   For a  first  cut, a  number  of
features can be eliminated to save you time, and me money.  I will enclose
his comments.

	I probably will be able to loan  you an LSI-11 when you are  ready
for it.  But, you will have to simulate the force balance and interface to
check your routines.   I can supply  typical data and  results so you  can
easily check yourself.

	The annual AI orgy/olympics  is on Friday.  I  will be at the  lab
then, so if you want to send a message, I will see it then.