[SIGCIS-Members] Message from Len Shustek re Stored Program

[Thomas Haigh]

[Len is also having posting problems with the list. While I'm here, to
address his final point, self modification of code goes much earlier than
Wilkes, Wheeler & Gill. The instruction set described original 1945 EDVAC
Report would have relied on code modification during execution for looping,
conditional branching, or indexing through a matrix. However this capability
was confined to address modification rather than unrestricted instruction
modification. This restriction was relaxed in subsequent work by von
Neumann's group and, as mentioned earlier, in Turing's 1946 description of
the ACE design. Designers of early stored program computers added
conditional branch instructions and indirect addressing capabilities such as
Manchester's famous index register to sharply reduce the need for self
modifying code. There's a clear description of this in Priestley's Science
of Operations, particularly pages 164- 172.] 

Tom,

I'm not sure why the first appearance of the precise bigram "stored program"
is so significant. There were certainly very early near-misses, such as this
from the 1947 Burks, Goldstine, and von Neumann "Preliminary Discussion"
paper:

"It is evident that the machine must be capable of storing in some manner
not only the digital information needed... but also the instructions which
govern the actual routine to be performed. ... Hence there must be some
organ capable of storing these program orders."

Even earlier is this from the 1945 "Description of the ENIAC and Comments on
Electronic Digital Computing Machines". It is farther from the precise
linguistic target, but exactly on point conceptually:

"This requirement as to instruction speed can be met in a direct and simple
fashion by storing instructions (in numerical code, so to speak) in memory
elements or registers of the same type as those used to store the numerical
data."

If your focus is on the ability of programs to modify themselves, I don't
know whether there is a reference earlier than the one in Wilkes, Wheeler,
and Gill that Allan Olley pointed out. The significance of that ability is
interesting to ponder. Doron Swade has done some research in this area, and
in his article in the December 2009 "EDSAC at 60" issue of The Computer
Journal he says of the expert opinions he received, "whether the primary
benefit was one of principle or practice was frustratingly blurred."

-- Len Shustek

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.sigcis.org/pipermail/members-sigcis.org/attachments/20120404/7725b612/attachment-0001.htm>