Forgot your password?
Cyhist Feb 9, 1997 B
Document Actions
Date: Sun, 9 Feb 1997 10:08:23 -0500
Reply-To: "CYHIST Community Memory: Discussion list on the History of
Cyberspace" <CYHIST@SJUVM.STJOHNS.EDU>
Sender: "CYHIST Community Memory: Discussion list on the History of
Cyberspace" <CYHIST@SJUVM.STJOHNS.EDU>
From: "A. Padgett Peterson P.E. Information Security"
<PADGETT@hobbes.orl.mmc.com>
Subject: CM> Microprocessors and Busicom
______________________________________________________________________
Community Memory: Discussion List on the History of Cyberspace ______________________________________________________________________
>Oh, so you *want* to hear a war story ;-) When i first started in telecommunications, when men were men, and modems ran at 300bps.
Well, in those antidiluvian daze when I started playing modems were two- man-lift rack mounts (by Collins) & channel speed was 66 wpm. Men were strong as bulls and smelled the same.
>The first of those programs had to be very concise, because of timing and the low volume RAM's ROM's available, later we got sloppy having 8K ROM and 16K RAM to play with :-)
Programs have always been either sloppy or elegant, that is human nature and the platform did not matter. The only difference was that in days of yore there were fewer machines and the demands were less. Have been "pushing the envelope" for over thirty years now and if anything the workload is increasing as the systems grow more complex and the underlying architecture gets ever sloppier (am fighting Office97 VBA5 right now and MS's record for abyssimal documentation has gotten worse and they seem to have forgotten to include some essential features for anti-viral use while translating many of the viruses automatically.)
>Is this accurate? How tulmultuous
>were these changes? Anyone out there who recalls what that transition was like?
Just like the changes we are seeing today. Hardly a ripple as the changes occured, just some companies who adopted them early were able to underbid others for the same thing. There was no sudden change for most people, just programmers stopped having to write microcode for every machine, they could an assembler which meant you did not have to calculate every jump or know the address of every variable. Relocateable code beame possible.
But the real root of the changes were memory. With 4k of available storage, you *had* to know exactly how everything fit together. With 32k you had more breathing room - for a while - by 1979, designing control programs to fit into 32k meant using a *lot* of overlays. This meant that memory space had to be partitioned between what remained resident and what was swapped - once designed a major program with 18k resident and 14k swapped. Each swap module wound up being 14k +0 - 100 bytes. Occasionally the same module would be present in more than one swap module because it turned out faster but it did not need to be resident. That was a real effort because each byte had to be counted. Today with 16Mb being U$100 and no real roof on the processor, why bother ?
Like most changes, were hardly noticed by the people involved at the time, were more concerned with being able to produce what the marketeers had promised at all so any advance was welcome.
Was also a time when we *expected* more/better/faster next year because it always happened. - Remember desinging a circuit once around a chip that did not exist at the time - a 16k static RAM, but three different manufacturers had released PDS on them & was willing to bet that but the time the boards were ready, one would be available in the quantities we needed. It was 8*).
Progress has mostly been problems that had no conventional solutions which forced engineers to seek the unconventional. In 1976-77 when I built the first 8080 based test stand for a gas turbine engine, it was not "pioneering", rather there was not enough time to build a conventional dedicated relay system so needed "something else" - turned out to be reusable and 1/5 the cost but was not the reason.
Similarly in 1973 when designed an infrared detection system with fiber-optic data link to reduce scrap in aluminum die casting machines, was not a matter of foresight, rather a problem that could not be solved conventionally. No one thought much about it at the time (though the union filed a grievance for "work-speed-up" and against me for bringing in a pencil iron when the smallest the electrician had was a 100 watter...). Do remember the thought process though - nothing electrical could survive the 1250 degree temeratures that needed to be measured except thermocouples and they were too slow. Needed something that could detect very fast hence the infrared, and needed to put the detector in a more benign environment hence the fibre-optic. Seemed logical at the time.
So there were no "paradigm shifts" any more than there are today when much more dramatic changes are taking place, rather a slow movement toward what worked that only looks dramatic when viewed in perspective.
Just like the transistor "revolution" - did not happen suddently - in consumer radios, the first appeared in 1954, did not achieve price parity until about 1957 (and used germanium transistors then which had a larger drop hence needed to dissipate more heat - real revolution was when Si replaced Ge but who noticed that ? And tube radios were still being made until 1963 - televisions in hybred form remained until the late '70s.
Am appalled at the schlock that passes for programming today yet can understand why. In the '60s-'70s were not many programs being developed and primarily for a small number of companies. Programs were command line oriented with occasional simple screen displays. Number or *real* programmers were probably in low four digits. Today the number of *companies* is in four or five digits and everyone is a programmer. Meanwhile the number of talented programmers has perhaps doubled. What do you expect ?
Warmly,
Padgett
______________________________________________________________________
Reply-To: "CYHIST Community Memory: Discussion list on the History of
Cyberspace" <CYHIST@SJUVM.STJOHNS.EDU>
Sender: "CYHIST Community Memory: Discussion list on the History of
Cyberspace" <CYHIST@SJUVM.STJOHNS.EDU>
From: "A. Padgett Peterson P.E. Information Security"
<PADGETT@hobbes.orl.mmc.com>
Subject: CM> Microprocessors and Busicom
______________________________________________________________________
Community Memory: Discussion List on the History of Cyberspace ______________________________________________________________________
>Oh, so you *want* to hear a war story ;-) When i first started in telecommunications, when men were men, and modems ran at 300bps.
Well, in those antidiluvian daze when I started playing modems were two- man-lift rack mounts (by Collins) & channel speed was 66 wpm. Men were strong as bulls and smelled the same.
>The first of those programs had to be very concise, because of timing and the low volume RAM's ROM's available, later we got sloppy having 8K ROM and 16K RAM to play with :-)
Programs have always been either sloppy or elegant, that is human nature and the platform did not matter. The only difference was that in days of yore there were fewer machines and the demands were less. Have been "pushing the envelope" for over thirty years now and if anything the workload is increasing as the systems grow more complex and the underlying architecture gets ever sloppier (am fighting Office97 VBA5 right now and MS's record for abyssimal documentation has gotten worse and they seem to have forgotten to include some essential features for anti-viral use while translating many of the viruses automatically.)
>Is this accurate? How tulmultuous
>were these changes? Anyone out there who recalls what that transition was like?
Just like the changes we are seeing today. Hardly a ripple as the changes occured, just some companies who adopted them early were able to underbid others for the same thing. There was no sudden change for most people, just programmers stopped having to write microcode for every machine, they could an assembler which meant you did not have to calculate every jump or know the address of every variable. Relocateable code beame possible.
But the real root of the changes were memory. With 4k of available storage, you *had* to know exactly how everything fit together. With 32k you had more breathing room - for a while - by 1979, designing control programs to fit into 32k meant using a *lot* of overlays. This meant that memory space had to be partitioned between what remained resident and what was swapped - once designed a major program with 18k resident and 14k swapped. Each swap module wound up being 14k +0 - 100 bytes. Occasionally the same module would be present in more than one swap module because it turned out faster but it did not need to be resident. That was a real effort because each byte had to be counted. Today with 16Mb being U$100 and no real roof on the processor, why bother ?
Like most changes, were hardly noticed by the people involved at the time, were more concerned with being able to produce what the marketeers had promised at all so any advance was welcome.
Was also a time when we *expected* more/better/faster next year because it always happened. - Remember desinging a circuit once around a chip that did not exist at the time - a 16k static RAM, but three different manufacturers had released PDS on them & was willing to bet that but the time the boards were ready, one would be available in the quantities we needed. It was 8*).
Progress has mostly been problems that had no conventional solutions which forced engineers to seek the unconventional. In 1976-77 when I built the first 8080 based test stand for a gas turbine engine, it was not "pioneering", rather there was not enough time to build a conventional dedicated relay system so needed "something else" - turned out to be reusable and 1/5 the cost but was not the reason.
Similarly in 1973 when designed an infrared detection system with fiber-optic data link to reduce scrap in aluminum die casting machines, was not a matter of foresight, rather a problem that could not be solved conventionally. No one thought much about it at the time (though the union filed a grievance for "work-speed-up" and against me for bringing in a pencil iron when the smallest the electrician had was a 100 watter...). Do remember the thought process though - nothing electrical could survive the 1250 degree temeratures that needed to be measured except thermocouples and they were too slow. Needed something that could detect very fast hence the infrared, and needed to put the detector in a more benign environment hence the fibre-optic. Seemed logical at the time.
So there were no "paradigm shifts" any more than there are today when much more dramatic changes are taking place, rather a slow movement toward what worked that only looks dramatic when viewed in perspective.
Just like the transistor "revolution" - did not happen suddently - in consumer radios, the first appeared in 1954, did not achieve price parity until about 1957 (and used germanium transistors then which had a larger drop hence needed to dissipate more heat - real revolution was when Si replaced Ge but who noticed that ? And tube radios were still being made until 1963 - televisions in hybred form remained until the late '70s.
Am appalled at the schlock that passes for programming today yet can understand why. In the '60s-'70s were not many programs being developed and primarily for a small number of companies. Programs were command line oriented with occasional simple screen displays. Number or *real* programmers were probably in low four digits. Today the number of *companies* is in four or five digits and everyone is a programmer. Meanwhile the number of talented programmers has perhaps doubled. What do you expect ?
Warmly,
Padgett
______________________________________________________________________
