Apollo Guidance Computer Part 25: Dan Lickly and Charles Simonyi Reflect on the AGC
Vložit
- čas přidán 4. 09. 2019
- A while ago, Mike, Ken, Carl and myself got invited at an evening talk at the Computer History Museum featuring two brilliant hosts: Dan Lickly, original AGC programmer, and Charles Simonyi of Xerox Alto, Microsoft Word and space tourism fame. This is my redux version of this fascinating event. Sorry for the coughs, I had a bit of a cold while recording this. My favorite quotes of the evening from Dan: explaining what the AGC did in the Apollo mission: “Everything!” and explaining scaling: "We were going to end up exactly half way to the Moon, or exactly twice as far".
Official CHM footage of the event: • CHM Live | If Software...
My website: www.curiousmarc.com
T-shirts on Teespring: teespring.com/stores/curiousm... - Věda a technologie
"if we don't get to the moon we are going to end up half way or twice as far....." Reference to floating-point math (or lack thereof) within the program. Fantastic!
I'm going to miss following along with such an ambitious project now that it's over
Excellent description by Mr. Lickly of the challenges they faced, problems solved. Intelligent and knowledgeable people like these can interweave humor into a technical discussion seamlessly. Excellent composition and editing by you as always.
I love how you cut to other clips to help us understand. Very good piece indeed.
Brilliant people who overcame the limitations of those early computers. We now have computer aided this and computer aided that but they had nothing but their amazing minds. Hats off to them, words can't express how much respect they deserve. In the future they will be looked upon as the geniuses they were.
i was interested to hear that dan worked on the polaris guidance system. my grandfather worked on the electromechanical parts of the same guidance system and was invited to work on the apollo program, but declined because he didn't want to work in florida again as he didn't like the weather there, and ended up going to work at ford aerospace instead. if he had decided to go work on the apollo program, i never would have been born.
These guys are legends but have the same geeky humor we all nerds have. Brilliant.
Dan Lickly still sharp as tac. It's amazing he still remembers all that.
It's almost unimaginable to most people today how precious and limited the computing capabilities were back in that time. Today, we walk around with multi-core networked supercomputers in our pockets, with orders of magnitude more computing capacity that was used to design nuclear weapons back in the day. But the cat videos look really great! The controller in your inkjet printer is closer in capacity to the AGC..
Thanks Marc. Was starting to get worried there wouldn't be more of these.
Nice to explain the asymmetric center of gravity for control. Thanks for such great content!
That freaked me out. Spin an asymmetric capsule around to steer it, that's nutso and brilliant at the same time .lol
I have a vague recollection of a comment, possibly by Cmdr Chris Hadfield, that Soyuz have a movable CoG, but it is entirely possible they use much the same as Apollo - fixed offset, and rotate capsule as required - which would seem more reliable than actually moving the mass. Although, moving the mass could allow faster and more refined adjustment.
It's good to see Don Knuth is still out and about. The last and only time I spoke to him was in 2002 at the University of Oslo.
I was waiting for so long for another episode of AGC series!! This made my day!!
Edit: After watching your till 0:45, I would like to tell you a thing: you and your team and Donald Knuth book's (also his interview) inspired me a lot to understand how computers work at the basic level and how one must write programs. Inspired by your SC/MP computer video, I started learning assembly language of AVR microcontrollers myself and just yesterday I made a little "kit" to program Atmega8A using toggle switches. I call it as Edith Mark I (imgur.com/a/HY3NtxN). I will make a better version of it, for sending 8 bytes for instruction is tough. Probably later I will work with programming 8085 in a similar fashion.
So, lots of _Thank You_ to you, your team and Donald Knuth.🙇🙇🙂
Breakfast with CuriosMarc, perfect 👌🏼
You bet
Such awesame discussions. Thanks for the diffusion of it. No other way to see those genius. Ha... excellent post-production Marc 😬👍👍👏👏
This series is so great.
Oh man. The part about Apollo 4 & 6. Talking about launching out of orbit then thrusting back towards Earth to reach lunar return velocities in order to test heat shields and stability. I never knew they did such tests during Apollo, but I absolutely did the same exact kind of tests preparing for missions in Kerbal Space Program. I'm not sure exactly what this means, but I really had a moment as he was describing those details.
Perfect for my morning coffee.
Thanks for sharing .
Hooray for a new CuriousMarc video!
Thankyou for this video :-)
Woah! At 2:29 you can hear MF tones bleeding across from an adjacent trunk line!
Thank you Marc :-)
What/where available is the portable AGC he was using right at the end of the video??
My guess is nowhere, probably a one off, but you can make your own, mikes AGC code is available, throw in some uC and some address logic for the display, set the component pads on a PCB with a hookup to the evaluation board plus the uC and the LED's and address logic, some simple power circuit, even an LM317 or so should probably be enough with a heatsink, autoroute it in eagle cad, clean it up, send off to a PCB fab house, order the selected components, wait a week, then build it. You can have the case and buttons 3D printed, and you're done. Have fun :-)
I am not sure where this one comes from, but the best I have seen are made and offered by Sam Ammons on ebay: www.ebay.com/i/264247811576. They are pricey though. They have a full linux distro inside running virtual AGC inside.
I bought the book!
3:37 Not a phrase you hear everyday...
A day off, breakfast in bed & Marc. What more do you need ?!!
You met Donald Knuth!! :)
You are so lucky.
0:50 Mike is melting
0.5 Hz clock speed? It’s amazing it could make all those reentry adjustments
That was the program cycle not the clock, iirc it was about 2Mhz?
Sort of. The oscillator worked with 2.048 MHz, which was divided down to 1.024 MHz to get a 4-phase clock. So AGC-CPU speed and todays CPU speed aren't really comparable.
Those 2 sec for one cycle Dan talks about is how often the AGC updates the state vector during powered flight, i.e. how often it reads the accelerometers in the stable platform. Reading them more often would've led to 'noisy' data and all that comes with it. Guidance itself is faster, the autopilot even more, it's triggered every 100 ms or so.
Just love this shit.
6:22 Wonder if Mr Simonyi is thinking "...now if only they had used my awesome Hungarian Notation..."
:)
2:28 Is there a purpose for these MF tones? They sound faster than the ones used in the telephone network.
I'm assuming you are referencing quindar tones. www.hq.nasa.gov/alsj/quindar.html
Sounds like crosstalk from an adjacent trunk. They used leased trunks between Houston and transceiver sites. I was pretty amazed to hear those!
There was no maths co processor you could plug into the AGC back then haha.
I watched the whole Video in the link.. VERY interesting but I hope hat there are more Videos being made or that the developers write down morre of the Story before it´s too late… what the Video did not Show us were the real struggles they had.. how they came to Solutions..why they decided to do it like this and not different.. I mean that for both Hardware and Software.. how close were they to give up.. were there any critical issues that almost made the Programm fail etc tec.. I mean thousands of People worked on that thing for years.. I`m sure there`s much much more to tell…
Who decided to go for a modular Computer... the Story About potting etc etc:
For me personally difficult to understand how so many People worked on that for so Long... what did a programmer or Hardware guy do every day :))
BTW, where are the AGCs today that flew the entry and returned to earth ?
19:46 Hahaha
Lickly jokes "what's reentry"" but it's unlikely he didn't know what reentry was after working on guidance for an IRBM.
No floating point.... yup, been there... Parallax's BS2 PStamp runs PBASIC - Parallax's BASIC interpreter - it does not do floating point so you have to scale things for integer math and scale them back when done. It's a real pain in the butt.
You normalized all your values based on the number of places past the decimal point (precision), typically 3. So if you had the constant Pi of 3.142, you made that constant 3142 in code (1000 times entered, not calculated). Then, if you needed to multiply it by 2, just multiply it by 2 to get 6284 and then divide by 1000 to get 6 (you lose the precision unless you left it scaled by 1000). If you needed to multiply Pi by 2.5, that would be 3142 X 2500 / 1000 to get 7855 which you could leave as is, treating the last 3 digits as being past the decimal point, or divide again by 1000 to get 7 (the integer value losing the factional part - there is no rounding).
The 1st real personal computer was not the Apple but the Altair 8800 by MITS.
I just don't know if we could do this now. It would be quite hard to do this right now.
Efficient programmers. Bitwise wizards.