The Apollo 11 RAM was less than a single modern email; your smartphone carries roughly two million times more working memory than the computer that landed humans on the Moon.
Apollo 11 RAM: The 3.5KB computer that landed on the Moon
To put this in perspective, a single high-resolution photo from a modern phone occupies more memory than the entire Apollo Guidance Computer had available for all real-time calculations. The AGC's RAM was not just small by today's standards; it was a deliberate engineering choice dictated by the weight, power, and radiation constraints of a 1960s spacecraft.
Core rope memory: The Apollo Guidance Computer's erasable storage
The AGC's RAM was built using magnetic core memory, a technology where tiny ferrite rings are threaded with wires. Each core could store one bit of data, with the direction of magnetization representing a 0 or a 1. This was the erasable storage: the computer could write new values to these cores during flight, which was essential for changing variables like velocity and altitude. The physical construction involved weaving wires through the cores, and the resulting memory was non-volatile, meaning it retained data even if power was interrupted. This was a critical feature for a mission where electrical glitches could not be allowed to erase navigation data.
The AGC also contained a much larger read-only memory (ROM), known as core rope memory. This ROM held 36,048 words of fixed software, including the operating system and the lunar landing programs. The distinction is important: the 36K words of ROM were permanent, physically woven patterns that could not be changed, while the 2K words of RAM were the scratchpad for all dynamic computations. The RAM was where the computer kept intermediate results, sensor readings, and the constantly updated state of the spacecraft's position and velocity.
How bank-switching squeezed performance from 2,048 words
With only 2,048 words of RAM, the AGC's engineers faced a severe constraint. The solution was a technique called bank-switching. The AGC had a total of 36,864 words of addressable memory, but the CPU could only directly access a small portion at any given moment. The memory was divided into banks, and the computer could switch between them by setting a register. This allowed the software to swap in different blocks of code and data as needed, effectively giving the AGC access to more storage than its physical RAM would suggest.
The bank-switching system allowed the computer to pull in the necessary guidance routines from ROM while using the 2,048 words of RAM for the live data that changed every second. The software was meticulously hand-written by programmers at MIT, who optimized every instruction to fit within the tiny memory footprint.
Memory specs vs. a modern smartphone
If you tried to run the Apollo 11 landing software on a modern phone, it would occupy a fraction of a percent of available memory, leaving the rest for apps, photos, and the operating system.
The phone in your pocket can hold billions of them. Yet with that minuscule 3.5KB of RAM, the AGC successfully guided a spacecraft to a precise landing on the Moon, processing real-time sensor data and executing complex guidance equations that would have been impossible without careful memory management.
The engineering lesson from Apollo 11 is not about raw capacity but about optimization. Every byte of RAM was accounted for, every subroutine was written to minimize memory usage, and every calculation was designed to leave enough room for the next one. The 2,048 words of RAM were not a limitation that hindered the mission; they were a constraint that forced the kind of disciplined programming that made the Moon landing possible. When you compare that to the gigabytes available in a modern device, it is a reminder that computational power is not just about hardware specifications but about the ingenuity of the people who write the software.
