رایانه Apollo Guidance ("AGC") برای پشتیبانی از فرود ماه ماه آپولو بین سالهای 1969 و 1972 از روی فضاپیمای آپولو استفاده شد. این گفتگو "همه چیز در مورد AGC" را شامل می شود ، از جمله طراحی سخت افزاری عجیب اما هوشمندانه ، سیستم عامل انقلابی آن و چگونگی نرم افزار آن به انسان ها اجازه می داد ماه را بپیمایند و اکتشاف کنند.
مایکل استیل
کریستین هسمان
If the presenters are too fast, play it at 0.75 speed or lower.
Absolutely amazing!! Enthralling and overwhelming
The best part was units conversion to Imperial units just for the sake of the astronauts.
Its sad this was only used seven times. Such a monument of genius for so little utility. If you told me I had to master this guidance computer in high school in order to become an candidate for the space program, I would've accepted the challenge. I wonder if any of the Apollo astronauts had nostalgic dreams of ever interacting with this computer if it meant they could fly to the moon again.
Fascinating! But I assume they meant KB, not KW, also its much nicer to listen to at a slightly slower playback speed, 0.75 worked for me.
USA
this is like straight up assembly… how to implement with Python in RasPi? (asking for a friend)
How the hell could they fit all of that in 4K ?!? xD
chief you speak to much in souin souin that's the mike which needs a filter pass low!
Great presentation. Having worked with architects in the early DG Nova days, I learned to appreciate the creative thinking and applications of the hardware engineers. It's incredible what we implemented in 4k of core memory back then.
Outstanding
I’m maintaining a shipping product that uses an MCU with almost an exact duplicate of this architecture – without old time quirks. Windowed registers (16 byte and 256-byte windows), 2-4K of RAM, 16-64k of ROM. Clock is 24x faster, and machine instructions are faster, but over the 20 years it’s been in use, I have developed most of the same approaches, including prioritized multitasking and task recovery. I have independently rediscovered the NEWTASK variable 🙂 They truly found very appropriate ways of designing such a complex system. Such constrained architectures are fun to work with. A bit of an art to it. Thankfully, no C for that anymore. Nowadays I use Rust. AGC code can be 1:1 machine-translated to that MCU. Only a few AGC instructions result in more than 2 assembly instructions on the MCU. Amazing. Good engineering never dies.
There’s a site of some Russian (I presume) troll “engineer” who goes into incredible detail to discredit this stuff. He makes so many mistakes in his “analysis” that I’m almost sure it was a paid job. Who’d have time to write so much nonsense… I’m especially amused by his “deconstruction” of the AGC software architecture. I know it works, because when you work within these constraints and need the last ounce of performance, you end up doing it the same way – there’s no other way really. Only identical design can be similarly performant.
AGC software has top half/bottom half interrupt handling split, just like every modern OS kernel, for crying out loud. There have been many software designs (many DOS drivers, lots of 16-bit Windows drivers, etc.) done without such split, with all the attendant latency and thread-safety problems.
Cooperative multitasking vastly reduces the fallout from task switching and makes it unnecessary, for the most part, to write threadsafe code. That was a huge time saver and must have saved them from a lot of bugs too. AGC software was a marvel of engineering, no doubt about that, and it was way ahead of its time in all the best ways. This wonderful presentation gives that “old” code all the respect it deserves.
Quite a few MCUs implement interrupt handler setup in terms of “hidden” micro programs, just like AGC. Some old ST designs from the 90s reused the ALU for timer counters, just like AGC. It’s all solid design, and a testament to the creativity and engineering acumen of the people who developed the AGC and it’s software.
Hint. set .75 😉 works for me lol
Unbelievably clever for the time.
I'm halfway through and have caught only one error so far: the AGC was based on only one type of digital IC, a dual 3-input NOR gate (not a NAND gate).
There's a certain beawabout these old machines. It is very quirky, but on the other hand, it is entirely possible to learn everything about it because it is so small, which makes it a lot more satisfying than modern computing in some ways.
This talk already gets into very much detail down to machine microcode and it's only one (densely filled) hour. Compare this to modern computers where there's layers and layers of libraries for everything. Let's say you want to draw a chart on your webpage. You add the d3.js library as a dependency, add some lines of code, and voila, there it is. But good luck trying to understand down to machine microcode level how the lines of the chart appear on the screen.
Speak faster !!!!! it's really too slow !!!!! ja ja ja ja
Great lecture made more understandable by playing it back at 0.75 x normal speed
Super cool!!!! Great presentation!!!
And here I am, making scripts to do stuff using Python. Man, low-level is so incredible hard to grasp compared to something like Python.
very in deepth
Talk faster!
Brilliant.
if this was the Olympics, they just won the gold with a perfect 10.
This was intense but super interesting
Great! Works even greater in 0,75 speed.
Ahh man turns out i an not alone , i honestly thought i was crazy for knowing what Verb , Noun translated to meant for something that has not flown for 50 years .
Alles abgelesen runtergerattert …schade dass deutsche sooo schlecht im Englischen sind…
"Data" is a bloody plural!
That basic blunder makes me not trust anything in this clip.
This computer represent geniality of human bean, congrat tô NASA and all that made this possible !!!!
I write XAP2+ and 8051 assembler (as well as C) for a living, and some of the similarities with the AGC are spooky. I always wondered where some of the stranger ideas in these processors came from.
But please stop talking about it being guys that developed the AGC. There were some women working on the AGC, Margaret Hamilton for example (look her up on wikipedia).
Great explanation, I really appreciate that!
AGC prototype 14 is now completely operational again. See youtube channel of CuriousMarc https://youtu.be/AhVKYDVZxWQ
Amazing! What a ride. Great presentation, kudos for clarity and information density.
Would it be possible to speak a little bit faster?
Great presentation, if a little speedy. Audio needs a narrow notch filter centered at 150Hz.
The best of the best of the best with honors!
At the time that the hardware of the AGC was decided (1964), the 1's complement was already obsolete and had been replaced with the 2's complement (the IBM computer of the Saturn rocket was using the 2's complement); so why use a notation which was already obsolete and slower than the 2's complement?
https://www.youtube.com/watch?v=yl4BcSD1NaM
3:12 Whoa, is that CuriousMarc? Currently enjoying his series where he rebuilds this insane machine. The man is a living legend.
Talk too fast ….i don t understand nothing
Pretty sure Apollo 11-14 (minus 13) went back to moon orbit and cicularised before docking. Unlike the animation here. Later Missions, probabaly 15-17 docked directly with the cm without first going to orbit. They didn't need that safety net feature anymore.
And of course it didn't control the launch. That was planned, but Von Braun's team insisted on the Saturn 5 having it's own Computer. Which came in handy on A12.
recently, a lot of money goes to maintain the myth of landing a man on the moon
great work