NASA Mars Perseverance Rover Uses Same PowerPC Chipset Found in 1998 G3 iMac

Blackstick said:
Anything launched into space has typically been in a multiple years long project, sometimes decade(s) long. It also usually costs a gazillion dollars.

What matters most is reliability and predictability... especially for something going millions (or billions) of miles away and not coming back that we can’t send astronauts to physically fix.

Most of the closed loop systems aboard aircraft, spacecraft, submarines, weapons and similar applications have very old, proven, but mundane processors and operating systems aboard... the last thing you want is “the new kid on the block” when the stakes are so high. Although parts of it are modernized as needed, the ECU in a typical car has the processing power no more than a Texas Instruments calculator... some of the software in Boeing and Airbus jets is virtually unchanged since the 1980s as well.
Click to expand...
PowerPC variants are used in several current automotive and aerospace control modules. And yes, the hardware and software in many airliners, corporate, and military aircraft is quite old but very reliable. The $10K USD Garmin GTN 650 GPS nav unit for small aircraft is less sophisticated than a iPhone 4 but extremely stable.

Power Architecture | NXP Semiconductors

NXP Power Architecture portfolio is designed for automotive, consumer, industrial, and networking applications development.
www.nxp.com
 
PlayUltimate said:
Like others, I have a Lime and a Tangerine in the attic. I guess it's time to start the next project.
Click to expand...
I had the limited edition smoke-colored version. I loved that thing —just recently parted with it. Back then, there was a clear distinction between Apple and everybody else.
 
Last edited:
jefhai said:
I think this article must be incorrect?
Perhaps it’s using the same instruction set as the PowerPC 750

I would imagine NASA would have some error correcting cache and bit protection in their processor...

So they wouldn’t just use the same old chip / design
Click to expand...

Exactly. I also think the article must be incorrect. Same basic instruction set, even probably the same internal architecture at the uppermost functional unit level, but then heavily re-engineered for radiation hardening.

People here talking about it needing to be “super reliable” are only seeing half the picture. Once out of Earth’s magnetosphere cosmic/solar radiation can randomly flip bits causing data and/or instruction corruption. A basic tactic in radiation hardening is to have multi-level redundancy where say 3 or more ”elements” all do the same thing and there is extra vote counting circuitry that, if all 3 elements don’t present the same results, will select the majority result. I used the term “element” because sometimes an element is an entire circuit module or functional unit on a chip but it can even get down to the level of individual bits where for instance a 64 bit register might have each bit implemented as a set of 3 (or more) voting bits. I don’t know if this processor goes to that level but potentially very thorough radiation hardening can add a lot of complexity and goes way beyond just screening regular parts for ultra high reliability because in space even a 100% reliable part can still have that pesky radiation randomly flipping bits and it has to be able to detect and correct those bit-flips.
 
ArtOfWarfare said:
This is just total nonsense from Old Space.

SpaceX has demonstrated that the correct way isn’t to get crazy expensive parts, but to just build several from commodity parts (at a lower price for all of them than the price of building one from commodity parts) and test them while iterating on the design in rapid succession.

This is how cars were built early on. It’s how the Wright brothers worked. It’s how actual progress gets made.

Starting with 30 year old parts and then taking over a decade to run a test rarely leads to a successful conclusion. See, for example, Boeing Starliner vs SpaceX’s Dragon, or SpaceX’s Starship vs the SLS.
Click to expand...
Huge difference between craft that is in constant state of development where failure is a learning opportunity and a craft that is launched and will never be touched again. Tested reliability is far more important than having the latest tech. What if they had used the Tesla MCU with memory write cycle issues? There isn’t a service center on Mars to pull the Perseverance into to have it swapped.

Max communication speed under optimal conditions is 2 megabits per second. Commands from NASA can take anywhere from 6 to 44 minutes from request to confirmation. All post processing of the data is done on Earth. There is plenty of computing power on Perseverance to complete its mission.
 
I never understood why they can't just take a commodity processor, wrap it in lead, and call it a day. There's probably less than $200,000 in rocket fuel required to accommodate the weight from the extra lead.
 
While based off the same processor in the iMac. The RAD750 is much slower. Running between 110 Mhz and 200 Mhz. I'd assume to get it to use less than 5W.

MJaP said:
Wow, wouldn't have thought they'd still be making them... or equivalents. They'll be using vacuum tubes next, or selotaping a Speak and Spell to a Big Trak: "Next spell martian"
Click to expand...

The CPU used in the Apple II is still being manufactured and used.

MauiPa said:
Maybe there is a valid reason for using old chip sets that maybe could have been alluded to by the author? Barring that, it just makes it sound like the NASA engineers are a bunch of old guys whose last innovation was what 20 years ago? Who says you can't teach old engineers new tricks?

So come on, what is the reason? anybody? Is it radiation or something?
Click to expand...

Besides reliability which has been hammered to death by now. Perseverance is built off Curiosity. As long as the CPU used in Curiosity could fulfill the mission requirements of Perseverance. They'll use it. You don't fix what isn't broken.

It reduces development time and cost. They already have proof it works reliably. As such there are less variables to worry about in a successful mission outcome. If they change the CPU. They basically have to start writing all the software and OS all over again.

As they can't just take an off the shelf CPU and use it. Another company (in this case BAE) has to then make it function in extreme environments and use very little power. When planning of Curiosity began in the early 2000's. The RAD750 was likely the latest and greatest radiation-hardened CPU available. At least the latest which could run in the low power requirements of a Mars mission.
 
thejadedmonkey said:
I never understood why they can't just take a commodity processor, wrap it in lead, and call it a day. There's probably less than $200,000 in rocket fuel required to accommodate the weight from the extra lead.
Click to expand...
Weight of the overall rocket and payload is always a concern, at least for as long as we are launching rockets out of the Earth's gravity well.

Once we get bases on the Moon (and elsewhere) and we can manufacture stuff at those bases and launch from those bases, then weight won't be as much of a concern.
 
data.gif
 
1: The design/build cycle for spacecraft designed to travel to other worlds and never come back is decades long. If you started out with the A 14 being the CPU of your project with today as day 1, by the time it got to the end of the test out cycle, it would be more than obsolete. If you went back and replaced components based on such criteria, your project would never fly.

2: This is all the processing power they need for the project. It doesn't have to run GTA, it just has to sorta run Tetris, or maybe Angry Birds. You don't rent an exotic sports car to move to your new apartment across town. You rent a U-Haul.

3: Reliability is THE name of the game. No one wants their BILLION dollar program to drop a deuce on the surface of a planet 138 million miles away for the sake of some esoteric number crunching crown. The RAD 750 is tested and proven reliable in the harshest conditions known to humans. Do they charge a lot for it? Sure. Because they can. No one else has something on par with the total package it provides (reliability, familiarity of instruction set, proven track record, ease of integration with known past proven reliable sub systems). This isn't the time for project managers to cut corners.

4: I trust the engineers at JPL more than I do the people posting on an internet message board to figure this stuff out. I mean, they DID get a rover to Mars that's doing some amazing stuff. If anyone here can top it, speak up.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Back
Top