Why Carmakers Can’t Transition to Newer Chips • Hacker News Board

Re: Why Carmakers Can’t Transition to Newer Chips

#201

Post by csense » Sat, Oct 02, 2021, 7:57 PM UTC

I always thought chips were dominated by fixed capital costs of fabs. If they were dominated by variable material cost of wafers, as the article seems to imply, it wouldn't make sense that we see 90nm microcontrollers that sell for $1 and a high-end 16nm PC CPU that sell for $1000.

So the question is, what's the reason that 90nm microcontroller sells for $1?

I'm trying, and failing, to figure out an economic model that explains the market dynamics we actually observe.

If building a new 90nm fab costs $billions, almost as much as building a new 16nm fab, why does the 90nm microcontroller sell for 0.1% of the price of the 16nm Xeon?

If building a new 90nm fab costs 0.1% as much as building a new 16nm fab, why can't existing chip companies, some startup or GM themselves spend $10's of millions building a fab that can unblock $100's of millions of product, and alleviate the shortage?

Re: Why Carmakers Can’t Transition to Newer Chips

#202

Post by KZerda » Sat, Oct 02, 2021, 7:57 PM UTC

baybal2 wrote: post #34

01100011 wrote: post #29

Earlier quoted context omitted.

As far as I can tell, Tesla plays fast and loose, treating their product like a manufacturer of consumer electronics and not a manufacturer of a dangerous and durable good. That obviously allows them to out-compete other auto manufacturers who are more aware of things like product liability. See the Toyota accelerator-gate lawsuits. From https://en.wikipedia.org/wiki/2009%E2%80%932011_Toyota_vehic... : > However, on…

It is very troubling development when doing something so trivial even requires a computer. That's a matter readily solved by a PID controller made of a few dozen 74XXX parts. Possibly made n-times redundant. Even a more fancy LP solver probably wouldn't need a fully fledged computer.

Okay, so you've designed it. Now, make sure it works with the dirty electrical signal of a typical car's power bus. Then make sure it fits in the space requirements of under the car's hood, or in the door, or any of the other tight places that these embedded systems go. Then make sure that it's verified to act over several years of vehicle life. Then, make sure your suppliers will guarantee you that they won't discontinue the part for at least 10-15 years because of legal requirements for spare parts. Just throwing discrete logic at the problem doesn't always help.

Re: Why Carmakers Can’t Transition to Newer Chips

#203

Post by noir_lord » Sat, Oct 02, 2021, 7:58 PM UTC

VLM wrote: post #187

kazinator wrote: post #148

Any discussion about why you can't transition to newer chips that doesn't mention the software is incomplete. Chips are not fungible in part because of software compatibility. OK, so you got a newer chip, and have redesigned the board and everything to fit. Now you have to get all the old firmware running on it and validate it. If the new chip isn't a 100% backwards compatible version of the old one, including all th…

In the long run we'll only use FPGAs and soft cores. In 2050 if your old 2035 Ford needs a new ECU you'll just stuck a somewhat newer and larger FPGA on it, and if the 2035 ECU needs three hardware I2C bus with clock stretching, one of which bus has to do 10 bit I2C addrs and it also needs three hardware PWM pins with 11 bit resolution and two 8051 cores running at 5.25 MHz each and two CANBUS, you (or more likely Fo…

I think you are probably right, economies of scale mean that using a ridiculously powerful modern process to emulate an old processor often makes sense.

The C64 mini runs a modern(ish) arm processor and an emulator to pretend to be a C64 - that's a machine that is literal multiple orders of magnitude more powerful than the original system.

Re: Why Carmakers Can’t Transition to Newer Chips

#204

Post by R0b0t1 » Sat, Oct 02, 2021, 7:59 PM UTC

acranox wrote: post #197

snthd wrote: post #133

Earlier quoted context omitted.

>40-60V system What makes that voltage a better trade off?

Another factor I haven't seen mentioned in the responses yet is safety to human bodies. You (or your kid) can stick a wet finger in a 12V DC charger socket in your car and not get an electric shock, including any of the exposed contacts under the hood, including the battery terminals. But once you're up to 40-60V, the risk of electric shock to humans is actually something that needs to be factored in.

You'll get a shock if wet slightly below 9V. It just stays on your skin. Voltage penetrates dry skin at around 50V, and this is the legal definition of high voltage.

Re: Why Carmakers Can’t Transition to Newer Chips

#205

Post by R0b0t1 » Sat, Oct 02, 2021, 8:00 PM UTC

csours wrote: post #195

snthd wrote: post #133

Earlier quoted context omitted.

>40-60V system What makes that voltage a better trade off?

Other comments pointed out why higher is better. The cap at 60v is somewhat arbitrary, but as others said, higher voltage is harder to switch, and above 60v DC it is easy to kill people.

Voltage levels sometimes arise from the geometry of semiconductors, capacitors, etc. 60V is a common cutoff, usually to give some slack to a 48V design voltage.

Re: Why Carmakers Can’t Transition to Newer Chips

#206

Post by YZF » Sat, Oct 02, 2021, 8:00 PM UTC

acranox wrote: post #197

snthd wrote: post #133

Earlier quoted context omitted.

>40-60V system What makes that voltage a better trade off?

Another factor I haven't seen mentioned in the responses yet is safety to human bodies. You (or your kid) can stick a wet finger in a 12V DC charger socket in your car and not get an electric shock, including any of the exposed contacts under the hood, including the battery terminals. But once you're up to 40-60V, the risk of electric shock to humans is actually something that needs to be factored in.

40V is still fine IIRC. I think 48V is the threshold.

Re: Why Carmakers Can’t Transition to Newer Chips

#207

Post by VLM » Sat, Oct 02, 2021, 8:01 PM UTC

nbernard wrote: post #68

But for the cost, couldn't (for instance) a 16nm fab produce 90nm chips? If not, how complex would it be to "port" an existing 90nm chip, to produce a 16nm revision? Impossible, or quite easy but not cost effective? From the POV of the car companies, could such revisions be used directly, or would they need to be qualified in the same fashion as new parts?

If you're bored you can go to mycmp.fr and check their process catalog and compare a typical 55nm run to a 160nm run. Its not quite as standardized as ordering PCBs over the internet.

Different metallization (you usually don't get to choose Al or Cu) will have different resistances. Generally the smaller processes will be faster so a design with no race conditions or metastability problems on 160 might not run reliably or at all on 55. Some processes are analog oriented so they'll guarantee up to 60 volts or more, if you're trying to design power devices, other processes are logic oriented and they might have a standard voltage of 2.5, 1.1 etc.

You can design something that'll run on a 55nm process and something that'll give similar performance on a 160nm process BUT they'll be different designs. I think the closest analogy would be changing processes is like changing manufacturing material. You can make a car piston out of steel or aluminum but you can almost never just swap materials in an existing assembly line.

Re: Why Carmakers Can’t Transition to Newer Chips

#208

Post by tullianus » Sat, Oct 02, 2021, 8:09 PM UTC

sandGorgon wrote: post #176

csours wrote: post #51

Disclaimer up front: I work for GM. I don't work on chips or components for my day job. What follows is solely my own opinion. Some things to emphasize: OEMs (GM, Ford, Toyota, VW, etc) do not design components, and they do not want to. They design specifications for components, and then get suppliers to bid. This is great for efficiency in established ecosystems, not great for agility. To my knowledge, GM did not ca…

> If you were going to design a new car electrical architecture from scratch today, you would have something like a 40-60V system with a centralized controller (or pair of controllers in a safety redundant configuration). Actually this point is something im curious about - it seems to be redundant to have a 3X redundant processor ..probably placed in different places on the car body for safety. Would that not be supe…

Most SpaceX hardware works this way: off-the-shelf processor, triply redundant, using voting to figure out if one is wrong.

Re: Why Carmakers Can’t Transition to Newer Chips

#209

Post by brianwawok » Sat, Oct 02, 2021, 8:10 PM UTC

oceanghost wrote: post #183

csours wrote: post #121

Earlier quoted context omitted.

Small company risk is failure to grow. Big company risk is every other failure. Tesla is valued as a growth/tech company. As long as people believe in their growth, they get more cash. I don't personally understand how Tesla hasn't had more problems with their apparently uncontrolled engineering changes. At least part of it is customer enthusiasm for the product papering over any drawbacks. It's not that GM can't mak…

> I don't personally understand how Tesla hasn't had more problems It's that nobody cares and the media doesn't cover it. A friend of mine-- his Tesla has broken down a dozen times and he still happily pre-ordered the cyber truck. My 10-year-old Honda has never had any work done to it other than maintenance-- but that doesnt make the news either.

And if we go for antidotes, my current Tesla has had 0 problems or recalls, my previous Honda had many many recalls and issues.

This is not a useful way to look at a product.

Re: Why Carmakers Can’t Transition to Newer Chips

#210

Post by coryrc » Sat, Oct 02, 2021, 8:13 PM UTC

pedrocr wrote: post #198

coryrc wrote: post #127

Earlier quoted context omitted.

No and about as expensive as having designed it in the first place. Given design costs are the largest portion of per-unit costs it's not a good investment.

Could you explain why not? Naively, having a much more dense process should allow automatic conversion. Going from 90nm to 16nm is 10x the density, that's a lot of margin for an automatic tool to use. Why doesn't that work?

It's more than just a size change and there are features besides transistors. Say you have a 100 fF capacitor; is that because that's the right value based on an external constraint (say, interacting with a crystal) or because it's matching the inductance of a long internal path? Because you adjust them differently based on circumstance. And your transistors with a specific load must still supply the same current as before so they can't shrink as much as ones for internal logic.

Also because the materials have changed the dialectic constant probably has and now the relative sizes of components need to adjust. And circuits are designed to minimize switching loss based on the old switching time and now they'll be wrong.

Oh, and the breakdown voltage of the new process can't handle the voltage many of these old circuits use IIRC.