Re: Computers before Information Theory
On Sep 10, 2017, at 12:22 PM, Noel Chiappa via cctalk wrote: >> From: Brent Hilpert > > ...snip... > >> When/what/who was the actual first assembler conceived or produced? > > Noel wrote: > A very good question indeed! Does anyone know? > > I have this bit set that one early computer assigned the opcodes to make > sense as single characters; e.g. the ADD instruction would have had opcode > 'A' (not in hex, this was before that). Alas, I can't find which one it was - > it's not the Pilot ACE; I checked, and that was always programmed direcly in > binary (by punching the program onto cards in binary, manually). > A machine that springs to mind is the original UNIVAC. A was add, D was divide, E for Extract, T for Transfer, U for Unconditional Transfer, etc. Worked well for most of the instructions. The LGP-30 had the advantage of recoding the Flexowriter keyboard/tape input for its 16 instructions. B: Bring A: Add S: Subtract M: Multiply (fractional) N: Multiply (integral) D: Divide H: Hold C: Clear Y: Store Address R: Return Address E: Extract U: Transfer Control T: Test I: Input P: Print Z: Stop
Re: Computers before Information Theory
> On Sep 10, 2017, at 1:22 PM, Noel Chiappa via cctalk > wrote: > >> From: Brent Hilpert > > ... > >> When/what/who was the actual first assembler conceived or produced? > > A very good question indeed! Does anyone know? I don't. But I can point to an early example of a quite primitive assembler, the one in the Electrologica X1 ROM code (you could call that a BIOS). That was around 1957, written by Dijkstra and described in detail in his Ph.D. thesis. It has somewhat symbolic instructions (digit/letter, the digit indicates the operation and the letter the register, e.g., "2A" for "load into A register"). And it has symbols, but only two characters long, so they were used to label blocks of data or code, with numeric offsets from there. paul
Re: Computers before Information Theory
On 09/10/2017 10:22 AM, Noel Chiappa via cctalk wrote: > A very good question indeed! Does anyone know? Wikipedia gives credit to Kathleen Booth in 1947 developing an assembler for the ARC2 at the UofL. EDSAC had one in 1949. Then there was SAP (Symbolic Assembly Program) and SOAP (Symbolic Optimizing Assembly Program for the IBM 650 (ca. 1954?). Let's just say it's one of those obvious applications, especially if you've ever spent time hand-assembling code. --Chuck
Re: Computers before Information Theory
> From: Brent Hilpert > I have wondered just how much influence the latent theory that was > around influenced the practical implementors of calculating machinery > ... > My impression is the implementors at the time arrived at stored-program > machines far more out of practical necessity than trying to enact, or > even being much aware of, the theory. I suspect we'll never know for sure, because there may have been subconcious stuff going on that even the people themselves were not aware of. E.g. it's common to hear that 'Babbage had no influence on modern computers'. But... Aiken was well aware of Babbage's work, and it's reasonable to think that he had it in the back of his mind when doing the SSEC. And Aiken and the SSEC were well known to the early computing pioneers, so there's a path from Babbage to modern computers. Similarly for Turing's work - Turing and von Neuman knew each other (their paths crossed on numerous occastions, starting at the IAS in the late 30's, when Turing went there), and there are numerous people who say he was very familiar with Turing's work. > When/what/who was the actual first assembler conceived or produced? A very good question indeed! Does anyone know? I have this bit set that one early computer assigned the opcodes to make sense as single characters; e.g. the ADD instruction would have had opcode 'A' (not in hex, this was before that). Alas, I can't find which one it was - it's not the Pilot ACE; I checked, and that was always programmed direcly in binary (by punching the program onto cards in binary, manually). > However even before the stored-program machines, the Colossus machines > (WWII) were more logic/symbol processors than numerical. Bit of both, really - they did statistical work on streams of characters. > Shannon was consulted during the design of SIGSALY, IIRC Turing was consulted, too - but I think more in a role of checking the work, rather than doing any himself. See "Enigma", pp. 246-248. > That whole era of Nyquist/Shannon looking at the nature of information, > Turing looking at highly abstract theory of symbol manipulation, and > the implementors of calculating machines, that all came together to > produce the modern computing and informatics world can be fascinating. There's an excellent book which covers some of this ground, "Turing's Cathedral", by George Dyson (son of Freeman). Noel
Re: Computers before Information Theory
On 2017-Sep-06, at 11:03 AM, Paul Koning via cctalk wrote: >> On Sep 6, 2017, at 1:25 PM, Fred Jan Kraan via cctalk >> wrote: >> >> While reading a biography of Claude Shannon, I try to get a picture how >> computers were seen and used before Information Theory emerged. It might be >> something like this: >> >> Before Information Theory, computers were mainly calculators; processing >> programs from numbers put into the machine, much like programmable, but >> non-graphic calculators. Information Theory states that almost any digital >> encoded data can be processed, as long as you can teach the computer how to >> interpret the data. >> >> Any more insights on this? > > It seems to me that Turing's 1936 paper clearly takes the information theory > route. Whether Babbage did I'm not sure. While there was theory around before the machines, the stored-program machines made it concrete. Once numbers and programs were stored in the same memory and hence represented the same way it "brought home" or made concrete, aspects of the theory. But that's not to suggest the theory was the driving force. I have wondered just how much influence the latent theory that was around influenced the practical implementors of calculating machinery in that era of the mid-1940s, the impression I have is not much, at least not in a recognisably causal or conscience manner. My impression is the implementors at the time arrived at stored-program machines far more out of practical necessity than trying to enact, or even being much aware of, the theory. Once those machines were implemented I think one of the first practical recognitions or uses of the data-vs-numbers, or symbols-vs-numbers, distinction would be assemblers. When/what/who was the actual first assembler conceived or produced? I think someone produced a graphical-display tic-tac-toe program for Whirlwind quite early, more a logic processing task than number-crunching. However even before the stored-program machines, the Colossus machines (WWII) were more logic/symbol processors than numerical. The SIGSALY digital voice encryption system (also WWII) was a mixture of digital numeric and logic processing for an objective that was not number-crunching. Shannon was consulted during the design of SIGSALY, IIRC from reading. But then also to keep in mind that Turing's work derives from Russell, et al reducing numbers and calculation to the more abstract notions of set theory, which - expressed in modern terms - was essentially declaring numbers and calculation to be a subset of 'information'. That whole era of Nyquist/Shannon looking at the nature of information, Turing looking at highly abstract theory of symbol manipulation, and the implementors of calculating machines, that all came together to produce the modern computing and informatics world can be fascinating.
Re: Computers before Information Theory
> On Sep 6, 2017, at 1:25 PM, Fred Jan Kraan via cctalk > wrote: > > While reading a biography of Claude Shannon, I try to get a picture how > computers were seen and used before Information Theory emerged. It might be > something like this: > > Before Information Theory, computers were mainly calculators; processing > programs from numbers put into the machine, much like programmable, but > non-graphic calculators. Information Theory states that almost any digital > encoded data can be processed, as long as you can teach the computer how to > interpret the data. > > Any more insights on this? It seems to me that Turing's 1936 paper clearly takes the information theory route. Whether Babbage did I'm not sure. paul
