Neil Armstrong landed on the moon 44 years ago with a 1.024 MHz computer. As the Google Lunar X Prize gears up, what kind of computing power will be used on man’s next mission to the moon?
This weekend marks the 44th anniversary of the Moon landing, unarguably the greatest technological achievement of the 20th century. This climax of this scientific and technological marvel was completed with one of the first uses of an integrated circuit — a crude modern processor — certainly a far cry from the computing power being used by today’s Lunar X Prize teams.
The Apollo Guidance Computer (AGC) was clocked at 1.024 MHz and had 74-kilobytes of available memory with 4-kilobytes available for what we know call Random Access Memory. The real achievement of the AGC was that it sported a real-time operating system, allowing the computer to run multiple tasks at once and prioritize tasks for CPU time given their priority.
The ability for the AGC to multi task stopped the mission from becoming a tragic failure during its final moments. During the final descent of Apollo 11’s lunar module, a hardware bug with one of the lunar module’s RADARs caused the AGC to needlessly compute data from the main RADAR in addition to the dedicated landing RADAR. This caused a series of alarms warning the crew and ground controllers that the processor was about to hit an executive overflow. However given the operating system’s ability to drop tasks based on priority, the operating system was able to free up enough processing power to complete the landing navigation program successfully.
Astronauts onboard Apollo 11 inputted compounds into the AGC using system called DSKY, which stood for display and keyboard. Commands were inputted with verb noun pairs that had numbers assigned to them, a sophisticated and elegant paradigm at the time in comparison to the punch cards regularly used in the computers of the day.
There are some advantages to using such a simple machine in spaceflight. The more complex a machine and its code, the greater likelihood of a bug arising. That’s why advanced jets today, like the B2 bomber, run on chips that are “Pentium Class” (the exact details are classified) in the era of Haswell.
To the moon, again
Trying to compare the computing power available today to that used in the Apollo mission is almost futile due to the vast differences in processing power.
Consider the Asimov rover being built by X Prize team “Part Time Scientists”. The team is using Nvidia Tesla GPUs — the clock speed measured in Gigaflops — on their workstations to simulate the launch of the rover from earth and its landing on the moon. They are also using GPUs on their ground control stations to make sense of the wealth of data collected by the Asimov’s sensors, and to decompress the high definition video and images captured by the on-board cameras.
Within the CPUs and GPUs that the Part Time Scientists team use is the legacy of the Apollo era. The first integrated circuit, the first crude processor, was developed for the Apollo moon landing, and if the engineers that built that circuit today were still alive they would no doubt be amazed at the advances in computing that have occurred since then.