We devote two chapters of ENIAC in Action to the planning and running of a series of nuclear Monte Carlo simulations on ENIAC from 1947 to 1950. The first set of simulations were run for Los Alamos during March and April 1948 using a program designed over the previous year.
For its time this was a remarkably complex program, featuring complex mathematical expressions, jumps, a subroutine called two places, numerous examples of conditional branching, and nested loops. To run it, ENIAC was converted to a new control system based, like that of other modern computers, on the one described in John von Neumann’s 1945 “First Draft of a Report on the EDVAC.” The spring 1948 Monte Carlo simulation was the first time code written in this form, which we call the modern code paradigm, was ever executed. Other computers had read programs one coded instruction at a time from paper tape, but without an addressable memory they did not perform automatic jumps or branches.
The most complete flow diagram we located for the original version of this program is dated December 9, 1947. It fills a desk when unfolded, and so could not be reproduced in full in the book. We believe the full diagram to be of considerable historical interest and have produced a poster version, annotated with some comments and a summary of the different main sections of the program. It also includes images of a draft diagram, a snippet of the ENIAC program code, and the first handwritten page of a report by Klara von Neumann. She coded the program and worked with John von Neumann and Nick Metropolis to design it.
The image above links to a high quality PDF, suitable for printing at poster size (18″x24″). Thanks to support from Mrs L.D. Rope’s Second Charitable Settlement we are also able to mail high quality printed versions free of charge to the first 100 people to request them. Requests should include a full mailing address and be sent to Thomas.firstname.lastname@example.org.