Silverfrost Forums

Welcome to our forums

FTN compiler used in 1979 paper

29 Jun 2013 6:07 #12534

While working through some papers on profile reduction I came across a paper dated 1979 where the FTN compiler was used.

In the first paragraph of section Results and Discussion (p. 848) this is mentioned. The paper is also on-line available: 1979 paper by Everstine

29 Jun 2013 9:57 #12535

I used a CDC6400 (when it was new) at Imperial College, London in the early 1970s and also in the late 1970s . Most of what I did made use of the MNF (Minnesota) Fortran compiler, which was Fortran 66). However I do remember reading the sparse documentation on a compiler named FTN. This had plotter graphics that made use of the Hershey character vectors - we could get output on Calcomp and Kingmatic plotters. I have long been convinced that FTN was an early precursor to today's FTN compiler, not least by the use of Hershey characters which were an important feature of FTN77/DBOS. (A CDC version doesn't appear in the history). I can't remember well, but the FTN I was shown might even have been a new-fangled Fortran 77. Anyway, I remember sticking to MNF for familiarity reasons, and not using Fortran 77 until I used a VAX.

E

1 Jul 2013 12:50 #12542

Salford University's FTN77 wasn't available on the CDC6400 - it was available on the ICL 1900 series: http://www.silverfrost.com/53/ftn77_personal_edition.aspx

The DBOS run-time came with Hershey characters -- I suppose from a similar source than the one you remember on the CDC6400.

1 Jul 2013 3:56 #12543

Yes, there are plenty of references on the web to CDC Fortran Extended, and FTN is probably the best 3 letter contraction of Fortran.

'Twas the Hershey characters that convinced me.

E

1 Jul 2013 7:09 #12546

Quoted from LitusSaxonicum Yes, there are plenty of references on the web to CDC Fortran Extended, and FTN is probably the best 3 letter contraction of Fortran. E

I used a CDC6600 as well as a CDC6400 in the early 1970s. The former sported four Fortran compilers, three of which came from CDC: the RUN, FUN and FTN compilers. The fastest was RUN, whereas FUN did some optimizations but ran reasonably fast. FTN was the slowest but produced (most of the time) the fastest object code. It is possible that the three were actually just a single compiler but with different default options.

Bitsavers has one of the RUN documents online (PDF): http://bitsavers.trailing-edge.com/pdf/cdc/cyber/cyber_70/chippewa/Chippewa_Fortran-Run_Apr66.pdf This document contains rather detailed descriptions of the structure and operation of the compiler. If you read this document, remember that the CDC had a word size of 60 bits, and a 6-bit character set (the contents of an 80-column card fitted into 10_8 words).

The fourth Fortran compiler was the student-oriented MNF compiler, which aimed for fast compilation and gave copious warnings, diagnostics and variable maps.http://www.comp.tc-ieee.org/archive/FORTRAN.html

Please login to reply.