34C3/presentation.md
c6cd706939cdmaster

presentation.md
View Options

			% 34C3recap
			% Jean-Baptiste Aubort; EPFL//R//SCITAS
			% \today

			# 34C3

			![](./res/34C3.png)

			# 34C3

			## Chaos Computer Club & Communication Congress

			* 34th Chaos Communication Congress in Leipzig
			* First time in Leipzig, had to leave Hamburg because the congress center was
			being rebuilt.
			* This year venue was HUGE !!!

			![](./res/venue.png)

			# 34C3

			![Congress is +15k visitors, +2.3k volunteers, 50k clubmate bottles](./res/congress.png)

			# 34C3

			## Network

			* Uplink: 400Gbit/s, peak usage: 42.2 Gbit/s
			* Peak wifi users: 7673, APs deployed: 179
			* But the network was sometime very shitty because of SSID spam for instance

			![SPAM](./res/ssid-spam.png)

			# 34C3

			## Network

			* Some strange patterns appeared on the network graphs.

			![strange patterns](./res/bandwith-morse.png)

			# 34C3

			## Network

			* Some strange patterns appeared on the network graphs.

			![34C3 in morse code](./res/bandwith-morse2.png)

			# 34C3

			## Science is broken

			* Generating random experient data and have significant results
			* Publication bias, only the significant results are published
			* Mostly subconscious. What can we do to prevent p-value hacking ?
			* Usually nothing is done
			* Replicating studies
			* Talk about failure
			* Decouple study and data gathering
			* Preregristration for publication (and later check that publication is in
			line with the methodology presented in registration) -> already the case
			in drug study
			* Create the protocol first then publish it before the experiment. Publish
			the result indenpendently of the outcome
			* Sharing (data, methods, code, ...)
			* Good results (or negative results) are often boring,
			less attractive to the press but closer to the truth
			-> TED Talk science VS Boring science ?

			* The presentor blog https://betterscience.org/

			# 34C3

			## The Ultimate Apollo Guidance Computer Talk

			* Presentation of the computer that was used to control the command and the lunar modules
			* It was more of a how computers works talk :-D It's computer rocket science !
			* The talk is segmented into 4 parts: Architecture, Hardware, Peripherals and Software

			![AGC](./res/agc.png)

			# 34C3

			## The Ultimate Apollo Guidance Computer Talk - What does it do ?

			* Maintain State Vector
			* Stabilize Attitude
			* Calculate Burns
			* Control Burns
			* Monitor/Control Launch

			![](./res/agc-system.png)

			# 34C3

			## The Ultimate Apollo Guidance Computer Talk - Architecture

			* Von Neumann architecture
			* 15 bit word + 1 parity bit
			* 36 instructions (performance vs complexity)
			* Single address space for code and data, 12 bit address
			* 1 accumulator register (most instructions use it)

			![Instructions set](./res/agc-instructions.png)

			# 34C3

			## The Ultimate Apollo Guidance Computer Talk - Hardware

			* 1 MHz clock
			* Microcode
			* Core (RW) and core rope (RO) memory

			![Hardware](./res/agc-hardware.png)

			# 34C3

			## The Ultimate Apollo Guidance Computer Talk - Peripherals

			* Gyroscope: monitor and control attitude (orientation)
			* Accelerometer: read acceleration on the 3 axis
			* Optics: measure relative position using celestial bodies, it can be moved
			* Landing radar: distance to the ground
			* Rendezvous radar: distance between lunar and service module
			* RCS jets: gas jets to modify attitude
			* DSKY: screen and keyboard
			* Communication

			![](./res/agc-system.png)

			# 34C3

			## The Ultimate Apollo Guidance Computer Talk - System software

			* Priority based: jobs queue is checked every 20ms, highest priority job is executed
			* Preemptive real-time: all the components must collaborate, timing is not garanteed (ex. interrupt)
			* Fault tolerant: Watchdog, recovery routine
			* Virtual machine: interpreted simplified set of instructions, can be mixed with native code
			* Drivers: a timer is fired every 120µs to read or write to every devices in a 8 stages sequence
			* Interactive: The pinball program is the shell of the agc. Mission control has a remote shell

			![](./res/agc-soft.png)

			# 34C3

			## The Ultimate Apollo Guidance Computer Talk - Mission software

			* [AGC operation, from liftoff to re-entry (click me, 12mn)](https://media.ccc.de/v/34c3-9064-the_ultimate_apollo_guidance_computer_talk#t=2746)
			* The complete software stack has been published in 2017 on Github in the Public Domain: https://github.com/chrislgarry/Apollo-11/
			* It's based on [scan of the original code on paper !](http://www.ibiblio.org/apollo/ScansForConversion/Luminary099/0040.jpg)
			* WCGW? job overflow, abort button got loose (overwrite abort program)

			![](./res/agc-mission-soft.png)

34C3/presentation.mdc6cd706939cdmaster

presentation.mdView Options

34C3/presentation.md
c6cd706939cdmaster

presentation.md
View Options