Computer with 1Hz cycle?

[Zatnikitelman]

I was trying to prove a point to someone on the internet about reaction speeds, and I thought about something I read a long time ago but am now having trouble placing. At one time, I thought I read that the Saturn V launch computer operated at a cycle of 1Hz because "even a Saturn V doesn't move that much in one second" but now I can't find anything even remotely close to it. At a second glance, I thought I might have read that when reading about Powered Explicit Guidance referring to the guidance loop frequency, but I can't that either. Does anyone have a clue what I'm thinking about here? I'm pretty sure the quoted line is right or at least close to it, but other than that, I just don't know.

Thanks.

 

[Izack]

I believe the Saturn V's guidance computer operated at around 2MHz, not 1Hz.

Source: [ame="http://en.wikipedia.org/wiki/Saturn_Launch_Vehicle_Digital_Computer"]Saturn Launch Vehicle Digital Computer - Wikipedia, the free encyclopedia[/ame]

I think it's been a long, long time since computers could only complete one cycle per second. 1Hz is very, very slow. Basic arithmetic would take multiple seconds per operation. A 1Hz guidance computer would have a 'reaction time' several times longer than that. Plenty of time for a Saturn V to spontaneously disassemble.

 

[Andy44]

I don't think any computer was ever that slow! Certainly no electronic computer.

I bet I could make a computer out of steam valves (A Babbage engine) that is faster than that.

1 cycle per second certainly sounds waaaaaay too slow to fly a rocket with.

Or did you mean 1khz or something?

 

[Linguofreak]

Many early computers could be clocked all the way down to 0 Hz (allowing an operator to manually step through a program instruction by instruction and examine register and memory contents on the front panel), but I'm not aware of any that had an actual normal operating frequency that low.

 

[Urwumpe]

There are also slower software cycles in every guidance software, like major and minor cycles for the PASS GPCs of the Space Shuttle. These can be very low like 0.5 Hz for some older computers... would need to look which multiple of 6.25 Hz was used by which cycle in the Shuttle again.

 

[Hielor]

1 Hz seems like it might be a doable clock rate for a Minecraft computer, since it's 10 redstone ticks...

 

[Artlav]

I think he meant the control loop cycle of 1Hz, not the CPU clock of 1 Hz.
That is, the sensor readings are received and the controls are computed once a second.

 

[Urwumpe]

Space Shuttle GPC:

Minor cycle is 40 ms long (25 Hz), Major cycle is 960 ms long (1.04 Hz)

(Can do 480,000 additions per second, more complex instructions need more CPU cycles)

Space Shuttle Main Engine Controller:

Minor cycle is 5 ms (200 Hz), Major Cycle is 20 ms (50 Hz)

(One addition took 2 µs on the initial controller, newer controllers had been faster)

The Gemini GDC has only 7000 instructions per second.

 

[Hlynkacg]

I was trying to prove a point to someone on the internet about reaction speeds, and I thought about something I read a long time ago but am now having trouble placing. At one time, I thought I read that the Saturn V launch computer operated at a cycle of 1Hz because "even a Saturn V doesn't move that much in one second"

I believe the quote in question was referring to the sampling rate for guidance/telemetry (which was around 1.5 Hz for the Apollo Guidance Computer) not the processor speed.

ETA:
:ninja:'d by Artlav

 

[BruceJohnJennerLawso]

Space Shuttle GPC:

Minor cycle is 40 ms long (25 Hz), Major cycle is 960 ms long (1.04 Hz)

(Can do 480,000 additions per second, more complex instructions need more CPU cycles)

Space Shuttle Main Engine Controller:

Minor cycle is 5 ms (200 Hz), Major Cycle is 20 ms (50 Hz)

(One addition took 2 µs on the initial controller, newer controllers had been faster)

The Gemini GDC has only 7000 instructions per second.

What is the difference between major and minor cycles?

 

[boogabooga]

I think the Manhattan Project "computer"- literally a room full of people who each performed a calculation over and over all day- could do better than 1 Hz.

 

[Urwumpe]

What is the difference between major and minor cycles?

Essentially just that a major cycle is composed of multiple minor cycles, and during each minor cycle you execute a set of different tasks or processes. Some important are for example executed every minor cycle, other tasks are only done every major cycle. The computer display is for example only updated twice per second.

Executing the tasks in cycles makes it possible with minimum technology to monitor how well the computer performs or if tasks are not executed in the time that you specified (keyword: real-time OS)

 

[BruceJohnJennerLawso]

Essentially just that a major cycle is composed of multiple minor cycles, and during each minor cycle you execute a set of different tasks or processes. Some important are for example executed every minor cycle, other tasks are only done every major cycle. The computer display is for example only updated twice per second.

Executing the tasks in cycles makes it possible with minimum technology to monitor how well the computer performs or if tasks are not executed in the time that you specified (keyword: real-time OS)

So, the idea here being that instructions are executed at given intervals with very reliable precision (not variable based on the resources availabe for the program at any given instant?)

 

[Urwumpe]

So, the idea here being that instructions are executed at given intervals with very reliable precision (not variable based on the resources availabe for the program at any given instant?)

Not instructions (those have a very different meaning in computer science), but yes. You have to think of tasks, a small computer programm, that is meant to be executed on a specified event. For example, a task can wait to be executed when a custom event is triggered (by changing an event variable in memory). Other tasks are meant to be executed every n microseconds or after n microseconds.

The programming language of the Space Shuttle is very powerful in that context, it has many special syntax elements for coordinating tasks.

 

[BruceJohnJennerLawso]

Not instructions (those have a very different meaning in computer science), but yes. You have to think of tasks, a small computer programm, that is meant to be executed on a specified event. For example, a task can wait to be executed when a custom event is triggered (by changing an event variable in memory). Other tasks are meant to be executed every n microseconds or after n microseconds.

The programming language of the Space Shuttle is very powerful in that context, it has many special syntax elements for coordinating tasks.

But the key idea here being that tasks are executed every n microseconds with great precision towards the value of n, ie the intervals at which tasks are executed are very regular, right?

 

[Urwumpe]

But the key idea here being that tasks are executed every n microseconds with great precision towards the value of n, ie the intervals at which tasks are executed are very regular, right?

Exactly. For such computers, you have a much higher requirement for task timing than on usual operating systems (Usually in the microsecond range). And much more important, while it does not really harm your windows PC, if one thread is delayed by a second, it is possibly an important error on a real-time computer.

You remember the infamous 1201 and 1202 errors on the Apollo 11 landing? That kind of error you get, when not every task can be executed in its scheduled cycle.

 

[Zatnikitelman]

I think Artlav nailed it, it was the update rate of the entire loop rather than the clock rate. Still, that seems very slow despite it being perfectly fine of course. If anyone has a link to a resource that indicates that, I'd love to see it to get better context.