MIPS in Space: Inside NASA’s New Horizons Mission to Pluto | Hacker News Reader

MIPS in Space: Inside NASA’s New Horizons Mission to Pluto

blog.imgtec.com

71 pointsalexvoica18 comments

The hardening argument seems disturbing to me giving the achievements that SpaceX have done in a short time. Apparently SpaceX is using some sort of redundancy system (https://lwn.net/Articles/540368/) to avoid all the hardening stuff enabling them to use more powerful and COTS hardware that NASA or ESA.

... or the COTS hardware is reliable enough in Low Earth Orbit.

Maybe... Anyway, it's an interesting question to be answered. If hardening technics could be improved significantly, it'd mean we could use incredible faster and cheaper hardware.

Low Earth Orbit is under the Van Allen belts, and so is shielded from a lot of the radiation of the inner solar system.

The outer solar system is dead quiet and quite safe... but of course you have to get there first.

doing a gravity assist at jupiter means you either have radiation hardening or you're cooked.

SpaceX seems to be using TMR. You'd have to look at the power requirements for TMR. According to TFA, power was a mission driver for New Horizons. Additionally, as others have pointed out, LEO is a very different radiation environment than deep space.

The hardening argument is hardly the most important one since it's a matter of having money to pay a silicon fab for the massive set up cost. According to [1] and [2], NASA has about as many still active deep space probes as there have been Falcon launches and many of them are as old or older than SpaceX the company, so we're talking about very different goals, histories, and organizations.

In order to achieve that kind of success while also dealing with the risk-averse public and obstructionist political system you need decades of reliability data for every mechanism, silicon fabrication process, and part in the space probe. Every cable tie, connector, all thermal and electrical insulation, every screw and panel, literally everything has to be extensively tested and their performance under a wide variety of conditions has to be completely mapped out.

For example, the last satellite I was involved in (NuStar, orbital xray telescope) which was designed mid to late 2000s, the PIs didn't even bother reviewing SSDs (Still at least a decade before they're used in critical missions) and hard drives were passed because no one wanted to risk our first xray telescope in decades, which happened to be several orders of magnitude more powerful than anything we've ever had in that spectrum. I left before the final decision was made but I have little doubt that the satellite uses tape drives fo rstorage because anything else is just not worth the risk.

[1] https://en.m.wikipedia.org/wiki/List_of_Falcon_9_and_Falcon_...

[2] https://en.m.wikipedia.org/wiki/List_of_active_Solar_System_...

Yes reliability and safety are the most important points but at same time, most of the certification processes to go to space are usually overbloated. Anyway, i'm just asking from the technical point of view if the radiation issue could be solved or not using some modern technics as SpaceX is using.

Overbloated - what set of requirements do you think are being unnecessarily exceeded?

NASA is a government agency overseen by elected representatives. Failure isn't merely a technical or even an economic setback - it can be a political risk as well. Every failed launch is ammunition for NASA's political adversaries.

Besides - this isn't a Falcon test capsule; it's a one-of-a-kind X-ray telescope. It's loss to the astronomy and astrophysics community couldn't be dismissed with a shrug of the shoulders and a quip on twitter.

The effect of radiation on a space craft/probe is not binary and like with all things, you have to take into account many different factors and make many trade offs often forced on you by economics and risk aversion/tolerance. In order to maintain public support and politic capital, NASA has to do some really stupid things that they wouldn't otherwise need to. SpaceX doesn't have to deal with public perception like NASA does because insurance companies and customers dealing with space are much smarter about the cost benefit.

For example (citation needed but these are ballpark numbers I remember from a JPL study), if you have $500 million for a mission like NuSTAR, one that must succeed because of its scientific significance, you have to spend that money to get a 95% success probability (for the primary mission duration). You can build and launch the satellite for $250 million with an 80% probability of success (and launch two) or launch one with a 95% chance for $500 million because that last 15% has rapidly diminishing returns on capital invested. The former has a 4% chance of both failing, 64% chance of both succeeding, and 32% chance of at least one succeeding.

However, after half a century NASA has found that what matters is the absolute number of failures, not number of successful missions or success rate. It's absurd, but NASA gets more funding when it successfully launches 4 out of 5 missions than if it launches 8 out of 10 for the exact same cost. This means that the two satellites for $250 million option has a one in three chance of at least one failure andh could easily cost NASA $500 million or so for another scientifically critical mission, even if one satellite makes it and the mission is successful.

It gets even worse when you factor in all of the pork barrel politics going into the NASA budget. Because NASA has operations and vendors in so many districts, in order for the timing and logistics to work out in the above example, they'd have to spend closer to $300-$400 million per satellite, wait many years between launch attempts, or build both satellites at the same time and waste everything but the launch cost (only $50-100 million) in the 80% scenario where the first satellite makes.

NASA has technical and political debt just like any fast moving 50+ year old organization but it is still a lean mean fighting machine that accomplishes milestones for our civilization on an annual basis. It's our political system that's bloated.

I love the table in the hardware section.

Thanks!

When I wrote my first article, a lot of people said "Why are they using old technology in a space probe?"

By creating that chart, I wanted to show that a space probe is not a phone or PC. You can't just throw octa-core SoCs inside, pack it up and send it flying into the night sky.

A digression, but it's amusing how far afield they had to go to find a modern SoC with a IMG GPU (vs. the rest of the world that is to first approximation all Qualcomm, with a few Mali's running around from Samsung). Obviously the real IMG flagship is the Apple A9, but that's not co-branded appropriately.

I wouldn't call Meizu, Xiaomi or other similar brands "far afield". They ship tens of millions of devices (and soon maybe even hundreds) every year.

But not on this particular SoC. Qualcomm (with Apple and Samsung not terribly close behind) remains the volume king even in China, last I checked. This other small-vendor SoC's are gaining momentum, but they're not there yet.

The reality today is that MediaTek is number two in mobile after Qualcomm. Meizu MX5 uses a MediaTek SoC as does the new Xiaomi Redmi Note 2 and many others.

Trust me, it's my job to know.

MIPS is a good choice here because the relatively small transistor count means it can be produced in a larger process, which gives better resistance to radiation.

Reply on news.ycombinator.com