PRINT(ID:5446/pri003)

Scientific calculation system 


PRINT - printing compiler for IBM 701 developed by Harroff and Fishman at GM Allison research lab, 1954. Part of a pair with READ

"...Don F. Harroff, James J. Fishman, and I wrote a pair of 'compilers' (now called macroprocessors) that turned out to be instrumental in implementing that project in a timely manner. My compiler, simply called READ, accepted specifications on the several data items in a record in decimal or alphabetical form and converted them to an appropriate internal binary form. The compiled program read data from the card reader and recorded them on magnetic tape at the fantastic density of 100 bits per inch! The mountains of input cards for Project X required only about 20 reels of magnetic tape.
Harroff and Fishman's compiler (called PRINT) did the inverse, converting data from binary to decimal or alphabetical form. The compiled program read data from magnetic tape and recorded it on the printer.
Special programs had to be written to perform any given analysis of this 'huge' data base, but needless to say it was very impressive to get an answer to a question in hours versus the days it took to do the same task on tabulating equipment."

Places Hardware:
Related languages
READ => PRINT   Co-development

References:
  • Ryckman George F. "The IBM 701 Computer at the General Motors Research Laboratories" pp210-212 view details Extract: SPEEDCODE and ACOM at GM Allison
    Most applications however, were programmed in SPEEDCODE or ACOM — two programming systems that transformed the single-address fixed-point arithmetic machine into a streamlined three-address floating-point system, SPEEDCODE was authored by Walter A. Ramshaw and his people at the United Aircraft Corporation. ACOM was written by Jack Horner and others at the Allison Division of GM. Both of these systems used subroutines to perform the floating-point arithmetic, which in turn slowed the 701 from its basic speed of 15,000 single-address fixed-point instructions per second to about 150 three-address floating-point instructions per second.
          in Annals of the History of Computing, 05(2) April-June 1983 IEEE (IBM 701 Issue) view details