FORTRAN inverter

Table 9.21 shows a simple FORTRAN computer source program INVERT, to invert 3x3 matrices. The source program is written in "Hollerith card image" format instructions must be between columns 7 and 72 a character in column 6 indicates that the line is a continuation of the previous line comment lines have "C" in column 1 (these are instructive but not essential). Table 9.13 shows the input data, and Table 9.14 shows the output data. Of course, the INVERT program would be much shorter in C+-1- ... 

Table 9.12 Fortran Program INVERT with Comments (FILE NAME INVERT.FOR)...

C---CALLED INVERTED.MX, TO BE WRITTEN AS FORTRAN LOGICAL 6 BELOW--... 

Table 9.13 Sample Input File "INVERT.FOR" "MYMX.DAT" for FORTRAN Program...

Table 9.14 Sample Output File "INVERT.MX" Produced by FORTRAN Program File "MYMX.DAT" "INVERT" from the Input...

Tables 9.18, 9.19, and 9.20 list the equivalent commands for several systems (all commands are followed by CRLF, "carriage return + line feed", or by pressing "RETURN" or "CR" on the terminal keyboard) these three tables deal with running the FORTRAN program INVERT discussed above. Finally, Table 9.21 gives some file management instructions on a UNIX system.

Table 9.20 Instructions for Running Program INVERT on a DEC 1077 Computer Using an IBM PC or Equivalent, with Digital Visual FORTRAN 5.0, Run in Disk Partition C Program FilesXDesigner Studio F90, and with This Partition Opened for Use...

In most intstances however merely being able to arrange the calculation as a series of vector operations, without worrying over the "unit address increment" requirement, makes extremely good if not maximal use of AFPP, VP or AP. As an illustration of this point Table XIII shows "normal" FORTRAN code for a pivotal condensation matrix inverter (the author is unfortunately by now anonymous) and Table XIV shows the vectorized version for the MVP-9500 at about two thirds completion. The VPLIB version is completely (as far as the author can manage at leastl) vectorized and written in assembler. Most of the vectorization is fairly obvious and only the reduction loops contain any obscurity. In order to maintain peak vector efficiency the MVP-9500 reduction loop does a little more work than is strictly necessary an alternative would ruin the vector flow. It is left as an exercise to the determined reader to unravel the full correspondence between Tables XIII and XIV. 

Scalar Fortran Coding of Gauss-Jordan Matrix Inverter... 

---