How do semiconductor companies plan for aging? There has never been a truly efficient solution, and according to this article, problems are getting worse. For example, electronics in cars continue to get more complex (and more safety critical). But cars are used in very different ways after being sold, and in very different climates.
When a device is used constantly in a heavy load model for aging, particular stress patterns exaggerate things. An Uber-like vehicle, whether fully automated or not, has a completely different use model than the standard family car that actually stays parked in a particular state a lot of the time, even though the electronics are always somewhat alive. There’s a completely different aging model and you can’t guard-band both cases correctly.
Aging is dealt with by heuristics, which typically add a “safety margin” to designs. But it’s not accurate, and leaves money (chip area = $ per chip) on the table.
Moreover, margin typically isn’t just one thing. It’s actually a stack.“The foundry, with the models that they give us, includes a little bit of padding to cover themselves,” said ANSYS’ Geada. “And then the library vendor adds a little bit of padding and nobody talks about what that is, but everybody adds up this stack of margin along the way. “
Source: Circuit Aging Becoming A Critical Consideration
But of course, the semicon industry has been dealing with emerging challenges like this for its entire existence. Each new problem starts at a low stage of knowledge, beginning with Stage 0 (nobody knows the problem exists) and usually ending at about Stage 6.
Here is an argument for allowing companies to maintain a lot of secrecy about how their data mining (AI) models work. The claime is that revealing information will put companies at a competitive disadvantage. Sorry, that is not enough of a reason. And it’s not actually true, as far as I can tell.
The first consideration when discussing transparency in AI should be data, the fuel that powers the algorithms. Because data is the foundation for all AI, it is valid to want to know where the data…
Source: The problem with ‘explainable AI’ | TechCrunch
Here is my response.
Your questions are good ones. But you seem to think that explainability cannot be achieved except by giving away all the work that led to the AI system. That is a straw man. Take deep systems, for example. The IP includes:
1) The training set of data
2) The core architecture of the network (number of layers etc)
3) The training procedures over time, including all the testing and tuning that went on.
4) The resulting system (weights, filters, transforms, etc).
5) HIgher-level “explanations,” whatever those may be. (For me, these might be a reduced-form model that is approximately linear, and can be interpreted.)
Revealing even #4 would be somewhat useful to competitors, but not decisive. The original developers will be able to update and refine their model, while people with only #4 will not. The same for any of the other elements.
I suspect the main fear about revealing this, at least among for-profit companies, is that it opens them up to second-guessing . For example, what do you want to bet that the systems now being used to suggest recidivism have bugs? Someone with enough expertise and $ might be able to make intelligent guesses about bugs, although I don’t see how they could prove them.
Sure, such criticism would make companies more cautious, and cost them money. And big companies might be better able to hide behind layers of lawyers and obfuscation. But those hypothetical problems are quite a distance in the future. Society deserves to, and should, do more to figure out where these systems have problems. Let’s allow some experiments, and even some different laws in different jurisdictions, to go forward for a few years. To prevent this is just trusting the self-appointed experts to do what is in everyone else’s best interests. We know that works poorly!
By now, Tesla’s manufacturing problems are completely predictable. See my explanation, after the break. At least Wall St. is starting to catch on.
Also in this article: Tesla’s gigafactory for batteries has very similar problems. That surprises me; I thought they had competent allies helping with batteries.
But one engineer who works there cautioned that the automated lines still can’t run at full capacity. “There’s no redundancy, so when one thing goes wrong, everything shuts down. And what’s really concerning are the quality issues.”
Source: Tesla employees say Gigafactory problems worse than known
My friend Don Norman wrote an op-ed this weekend calling for an FDA-like testing program before autonomous cars are put on the roads in the US. Clearly, some level of government approval is important. But I see lots of problems with using drug testing (FDA = Food and Drug Administration) as a model.
Here is an excerpt from a recent article about testing problems with Uber cars, which were the ones in the recent fatal accident. After the break, my assessment of how to test such cars before they are allowed on American roads.
Waymo, formerly the self-driving car project of Google, said that in tests on roads in California last year, its cars went an average of nearly 5,600 miles before the driver had to take control from the computer to steer out of trouble. As of March, Uber was struggling to meet its target of 13 miles per “intervention” in Arizona, according to 100 pages of company documents obtained by The New York Times and two people familiar with the company’s operations in the Phoenix area but not permitted to speak publicly about it.Yet Uber’s test drivers were being asked to do more — going on solo runs when they had worked in pairs.And there also was pressure to live up to a goal to offer a driverless car service by the end of the year and to impress top executives.
So Uber car performance was more than 100 times worse than Waymo cars?!
Academia has a problem: the value, necessity, and practices of collaboration are increasing, but the system of giving credit is inadequate. In most fields, there are only 4 levels of credit:
- None at all
- “Our thanks to Jill for sharing her data.” (a note of thanks)
- First Authorship (This is ambiguous: it may be alphabetical.)
- Listed as another author
In contrast to this paucity, modern empirical paper writing has many roles. Here are a dozen roles. Not all of them are important on a single paper, but each of them is important in some papers.
- Intellectual leadership.
- Source of the original idea
- Doing the writing
- Writing various parts, e.g. literature review
- Doing the grunt work on the stat analysis. (Writing and running the R code)
- Doing the grunt work of finalizing for publication. (Much easier than it used to be!)
- Dealing with revisions, exchanges with editors, etc.
- Source of the data.
- Raised the funding;
- Runs the lab where the authors are employed
- Source of the money: usually an agency or foundation, but sometimes the contracting author is listed as a coauthor.
It has always been clear that Musk does not understand high-volume manufacturing. Building rockets is very hard, but building 100,000 cars is very hard for a different reason! His predicted ramp rate was absurd. In the last 6 months, I think he has started to realize this.
Tesla has little chance of hitting its 5,000 weekly output during the fourth quarter. The chief reason: Its current production line can’t build vehicles at that rate unless it runs two 10-hour shifts seven days a week, which is
Source: Tesla | Hiccups Threaten to Slow Model 3 Launch | Industry content from WardsAuto
According to the article, Tesla has also deliberately ignored much of the accumulated wisdom about how to ramp in auto production. That might be OK for his second high-volume vehicle.
More details on Tesla’s ramp plans:
Tesla now has 2 choices, both bad:
- Go ahead and start building and shipping as fast as possible. The result will be multiple problems that require expensive hardware recalls.
- Add another 6? months to the schedule to run the as a pilot line for learning, rather than for volume. Expect zero salable output during that period. (As one of the comments said, they can give/sell those cars to employees.)
Added December 27: Tesla “still in manufacturing hell.”
Latest of many articles about Tesla manufacturing problems.
Here is a comment on that article: Musk needs to face up to having made a MAJOR mistake when he skipped some steps in the original manufacturing ramp-up.
He is probably also making another major mistake at present: adding new machines to the manufacturing process, before he has the existing machines working perfectly. This seems logical to people with no manufacturing experience, but it does not work. For one thing, it diverts his key resource, which right now is manufacturing engineers.
According to the author, naval ship handling still relies heavily on craft expertise. His article writes down some formulas and procedures to reduce collision risk. Source: When Am I Committed to Collision? | U.S. Naval Institute My own reaction is in a brief comment at the end of the article.
Here is another article in the same issue of US Naval Institute Proceedings that does a great job of explaining how collisions can happen, and why the captain of a USN ship is always responsible, and never completely safe.
This is the burden of command. A captain puts the lives of several hundred sailors into the hands of a young officer, typically 25 years old and typically green. So what does a captain count on to prevent disaster? The captain has “standing orders.” These are the rules in his or her ship that everyone (especially the OOD) lives by. …”