How Software Is Eating the Car (ieee.org) 281
The trend toward self-driving and electric vehicles will add hundreds of millions of lines of code to cars. Can the auto industry cope? From a report: Ten years ago, only premium cars contained 100 microprocessor-based electronic control units (ECUs) networked throughout the body of a car, executing 100 million lines of code or more. Today, high-end cars like the BMW 7-series with advanced technology like advanced driver-assist systems (ADAS) may contain 150 ECUs or more, while pick-up trucks like Ford's F-150 top 150 million lines of code. Even low-end vehicles are quickly approaching 100 ECUs and 100 million of lines of code as more features that were once considered luxury options, such as adaptive cruise control and automatic emergency braking, are becoming standard.
Vard Antinyan, a software quality expert at Volvo Cars who has written extensively about software and system complexity, explains that as of 2020, "Volvo has a superset of about 120 ECUs from which it selects to create a system architecture present within every Volvo vehicle. Altogether, they comprise a total of 100 million lines of source code." This source code, Antinyan says, "contains 10 million conditional statements as well as 3 million functions, which are invoked some 30 million places in the source code." How much and what types of software resides in each ECU varies greatly, depending on, among other things, the computing capability of the ECU, the functions the ECU controls, the internal and external information and communications required to be processed and whether they are event or time triggered, along with mandated safety and other regulatory requirements. Over the past decade, more ECU software has been dedicated to ensuring operational quality, reliability, safety and security.
"The amount of software written to detect misbehavior to ensure quality and safety is increasing," says Nico Hartmann, Vice President of ZF's Software Solutions & Global Software Center at ZF Friedrichshafen AG, one of the world's largest suppliers of automotive components. Where perhaps a third of an ECU's software was dedicated to ensuring quality operations ten years ago, it is now often more than half or more, especially in safety critical systems, Hartmann states. Which ECUs and associated software end up going into a Volvo like its luxury SUV XC90 model, which has approximately 110 ECUs, depends on several factors. Volvo, like all auto manufacturers, has variants of each model offered for sale aimed at different market segments.
Vard Antinyan, a software quality expert at Volvo Cars who has written extensively about software and system complexity, explains that as of 2020, "Volvo has a superset of about 120 ECUs from which it selects to create a system architecture present within every Volvo vehicle. Altogether, they comprise a total of 100 million lines of source code." This source code, Antinyan says, "contains 10 million conditional statements as well as 3 million functions, which are invoked some 30 million places in the source code." How much and what types of software resides in each ECU varies greatly, depending on, among other things, the computing capability of the ECU, the functions the ECU controls, the internal and external information and communications required to be processed and whether they are event or time triggered, along with mandated safety and other regulatory requirements. Over the past decade, more ECU software has been dedicated to ensuring operational quality, reliability, safety and security.
"The amount of software written to detect misbehavior to ensure quality and safety is increasing," says Nico Hartmann, Vice President of ZF's Software Solutions & Global Software Center at ZF Friedrichshafen AG, one of the world's largest suppliers of automotive components. Where perhaps a third of an ECU's software was dedicated to ensuring quality operations ten years ago, it is now often more than half or more, especially in safety critical systems, Hartmann states. Which ECUs and associated software end up going into a Volvo like its luxury SUV XC90 model, which has approximately 110 ECUs, depends on several factors. Volvo, like all auto manufacturers, has variants of each model offered for sale aimed at different market segments.
Comment removed (Score:5, Insightful)
Re:Cars without code? (Score:5, Interesting)
If you get something pre-1978 then the most sophisticated electronics in the vehicle will probably be the radio kit.
=Smidge=
Re:Cars without code? (Score:5, Interesting)
Re:Cars without code? (Score:5, Funny)
Depends on the car, for example the 1976 and later Cordoba which had The electronic Lean-Burn System which was probably one of the earliest examples of computer control on a Car http://www.allaboutvaliants.co... [allaboutvaliants.com] I just remember it being a total mess as my parents had one of these cars and an issue with the computer caused the engine to catch fire and, the elderly man who lived across the street from us running over with a fire extinguisher to put the flames out before things got bad
I had a 1965 Buick that caught fire as well, but it didn't take a computer to do that.
Re:Cars without code? (Score:4, Funny)
Re: (Score:2)
If you get something pre-1978 then the most sophisticated electronics in the vehicle will probably be the radio kit.
=Smidge=
I know quite a few cars in the early 70's had electronic ignition.
Re: (Score:3)
Re: (Score:3)
I had an early 1970s BMW 3.0 S. Carbs, not fuel injection. Can't remember if the ignition was electronic, but it did have a rudimentary servicing computer that told you when oil changes were due. Damn thing had a backup battery that leaked all over the board and ruined it. Didn't stop the car from running.
I miss that car - that thing could really lift its skirts and RUN.
Re:Cars without code? (Score:5, Interesting)
Yes they did, but that was all analog electronics, no microprocessor.
Right, but poster mentioned the radio, so I mentioned the ignition as another bit of electronic kit.
Might as well post this here:
The problem with the whole back to the good old cars is that as some one who grew up with them, they really weren't that good. I've pointed out a few eras for people who want one; some of the older cars are undeniably cool. I loved my old 65 Buick Skylark - it was a real sleeper - fun to challenge people with to drag race in my home town. They'd laugh until they could read my license plate crossing the finish line.
But, taking off the nostalgia goggles, that poor thing was finished, well before 100 Kmiles.
And lest people claim I abused the engine - a lot more went bad, and I didn't race very often.
Modern engines are truly wonders, pretty much the equivalent of blueprinted engines back in the day. The ECU's allow the engine to run more efficiently, ABS systems are getting better all the time. Traction control like my Renegade has, will tailor itself to what I'm driving in, mud, snow, whatever. Put it on auto, and it decides when to shift to four wheel drive. It's 9 speed transmission is computer controlled - although I can take manual control if I feel like it. But they figured that some might have concerns about some stuff, so it has an analog speedo/tach as well as a digital one. Old school knobs and touch screen for the radio of the number of temps and pressures I can call up.
The old school is best might claim that just makes more things to fail. But really, when you replace your points, plugs distributer cap and wires, and have a constant battle against rust of the old cars - I'm not sure that the complexity makes failure is a great argument.
Re:Cars without code? (Score:5, Interesting)
Of course it's all becoming a moot point since an all-electric vehicle has to have a considerable amount of electronics at minimum just to operate: battery bank management (including regenerative braking), and controller(s) for brushless drive motor(s). That's quite a bit right there.
Re: (Score:3)
We have better materials technology and better overall mechanical engineering now than we did, and we don't have a Detriot-dominated auto market that intends for you to buy a brand new car every 3 years -- hence the 100,000 mile lifespan before it falls apart. We also have synthetic engine oil that allows you to go 20,000 miles between oil changes. You could easily build a car or pickup today that has the longevity but without the massive amounts of electronic bloat: no 'infotainment' systems. Just an engine control computer and a stereo. Sadly, that's not what 'people' seem to want though. It makes me wonder how much cheaper something like a light pickup would be if you weren't forced to pay for all that.
A really interesting question!. It would take some modifications of the engine hardware. But the vehicle light busses would have to be replaced with a completely different wiring harness. Fuel milage standards would be altered by a lot. I was shocked to find out that my engine essentially shuts the injectors and spark down if it senses me going downhill for more than a few seconds. Pollution standards would almost certainly be affected as well. We don't meet them with the crappy stuff we used before.
My
1978 Ford Fiesta (Score:3)
I can sell you my 1978 Ford Fiesta parked in my dad's barn. You will have to tow it and do some repairs -- the fluid leaked out of the clutch pedal actuation so if you start it, you can't get the car in gear.
Seriously, by the 1980's engine control ECUs were a thing, but we are talking about the time of the original IBM PC with a 4.7 MHz processor, 256 kbytes of RAM, and dual floppy disk drives.
I am still surprised by the complexity of current cars. I thought that military aircraft and certainly space
Re:1978 Ford Fiesta (Score:5, Informative)
Spacecraft need rad hardening. Aircraft -- at least commercial aviation, I think military ends up with different rules -- have safety requirements that mean single-event upsets and similar failures cannot cause loss of life. And "poorly developed software has a bug" is definitely that kind of failure -- you need to satisfy RTCA/DO-178 (or the equivalent EUROCAE guidance) DAL A or B, depending on the unmitigated safety effect.
The slow pace of that development is mostly not code reviews. It's making sure the requirements are clear, complete, understood by everyone relevant, allocated to the correct components, and traceable to higher level requirements and to verification procedures. It's making sure the design documentation is clear, complete, understood by everyone relevant, etc. It's doing safety analyses to identify all the plausible failure modes and external causes of failure, what kind of failure detection and/or handling occurs, and what the mitigated and unmitigated safety effects are. It's doing enough verification to be confident that both the nominal and failure cases perform as expected. And, yes, it's also code reviews. But mostly it's those other things.
(If you look at something like the 737 MAX MCAS, you can identify development process failures in almost every one of those activities.)
737 MAX (Score:2)
Do we even know how the MCAS came to be? Have the accident investigations into the two disasters been completed and published yet?
I guess we know about the history of the design compromises with the Challenger and Roger Boisjoly and other engineers lobbying not to launch on that cold day and being overruled by both their managers and by NASA and about Richard Feynman getting encouragement "from a source" to do that public demonstration of dunking O-ring material in the ice water served to the Commission
Re:737 MAX (Score:5, Informative)
MCAS came to be because of a fundamental change in the feel of the 737. The MCAS system is only engaged during very few phases of flight - it is not needed when the autopilot is in use, which is pretty much 99% of the flight. Once you take off, you engage the autopilot. MCAS is disabled because the autopilot is flying the plane. MCAS only comes in for manual flight, and only because it helps restore the feel of the 737 Max to that of a 737.
Pilots fly it and make sure it "feels" right and familiar.
It's one advantage Airbus has because Airbus is true fly-by-wire, and the computer model enabling this can make the aircraft "feel" like any aircraft they want it to be.
Now, the big problem isn't the engineering that went into MCAS, but the management - they wanted a plane pushed out to compete with the A320neo ("new engine option" - which is why the 737 Max has new engines). MCAS isn't new, Boeing actually used it on a military plane years earlier to ensure plane stability. Of course, as anyone knows management wants it done yesterday and to hell with things.
What likely happened is Boeing management, being bean counters from Lockheed and not engineers, saw that the 737 Max was cobbled together from parts Boeing already had, so it shouldn't take long and forced it out the door much too early.
Boeing's problem is management, which is why it was referred to as "Lockheed taking over Boeing using Boeing's own money". They drummed out the existing Boeing management (who were engineers and closely monitored the engineering of aircraft), then promptly moved HQ from Seattle to Chicago. Boeing has no presence in Chicago - no facilities or anything. So now management is disengaged from the engineers, and management is just MBA types with no understanding of aircraft construction. Why else would they isolate themselves from the engineering?
The 737 Max is less computerized than modern planes. MCAS is just a compensation routine that is supposed to only apply a little pitch trim to make the plane feel like any other 737. The 737 is still a mechanically rigged airplane with power assist.
Re: (Score:3)
The Seattle Times [seattletimes.com] published a good high-level summary of the process failures for MCAS. This article [theaircurrent.com] goes into more detail about the emphasis on minimizing the re-training that pilots needed.
Boeing did safety analysis on an early version of the design, and submitted that to FAA as is required. They changed the requirements after doing that, and did not submit the revised analysis to FAA. There was a mismatch between the implementation and the system safety assessment (SSA, a term of art) about whether MC
Re: (Score:2)
The slow pace of that development is mostly not code reviews. It's making sure the requirements are clear, complete, understood by everyone relevant, allocated to the correct components, and traceable to higher level requirements and to verification procedures. It's making sure the design documentation is clear, complete, understood by everyone relevant, etc. It's doing safety analyses to identify all the plausible failure modes and external causes of failure, what kind of failure detection and/or handling occurs, and what the mitigated and unmitigated safety effects are. It's doing enough verification to be confident that both the nominal and failure cases perform as expected. And, yes, it's also code reviews. But mostly it's those other things.
Wow, is that what RTCA/DO-178 does?
Re: (Score:2)
I hope you like a lot of pollution with your simplicity, and breakdowns, and repairs.
Re: (Score:2)
Re: (Score:2)
Sorry. I'm unfamiliar with these breakdowns and repairs you speak of. Each Toyota from the mid-80s to the late-90's that we've owned (2 Camry's, 1 Corolla, and 1 small pickup) have easily given us over 200,000 problem-free miles over a 10-15 year period.
Remember the old Toyota commercials from the '80s calling on Toyota owners to tell them about their 200,000, and 300,000 mile Toyota's?
https://www.youtube.com/watch?v=kJJjtXS9_n0
They were simple and reliable back then.
Re: (Score:2)
I hope you like a lot of pollution with your simplicity, and breakdowns, and repairs.
No worries, just go electric [makesthatdidntmakeit.com] ;)
Re:Cars without code? (Score:5, Interesting)
This article actually has it backwards.
Yes, for a long time the trend has been towards increasing ECUs, based on the design philosophy of, "If you want a new feature, you buy a box from a Tier 1 that provides the feature, and wire it in." As a general rule, automakers love outsourcing work; for most of them, their dream scenario is that everyone else does all the work for them and they just slap a badge on it and take a cut.
The first company to break with this philosophy was Tesla, which has from the beginning had a strong philosophy of in-house software design, and built basically a "car OS" that offloads most vehicle software functionality into a handful of computers (with redundancy on safety-critical functionality). In recent years, other manufacturers have started to see the importance of this and are trying to clone Tesla's approach. Of "western companies", VW is probably the furthest along in this regards, although they hit a lot of snags (the ID.3 and ID.4 were supposed to be the first cars to use the new software stack, but were delayed because the software was unusable, and the first cars were delivered with a lot of bugs awaiting fixes). It's starting to mature up, though, and over the coming years one can expect to see ECU counts in VW vehicles continue to fall.
Eventually everyone is going to have to either make their own "car OS" stack or lease one from someone else. The benefits are just too significant. Lower hardware costs, lower assembly costs, lower power consumption, simpler cheaper lighter wiring harness, faster iteration time on new functionality, closer integration between different subsystems, you name it. This trend throws into reverse the notion of ever-increasing numbers of ECUs (which quite simply was an unsustainable trend).
Re:Cars without code? (Score:5, Interesting)
Yes, for a long time the trend has been towards increasing ECUs, based on the design philosophy of, "If you want a new feature, you buy a box from a Tier 1 that provides the feature, and wire it in." As a general rule, automakers love outsourcing work; for most of them, their dream scenario is that everyone else does all the work for them and they just slap a badge on it and take a cut.
It is not that simple. Take for instance someone like ZF quoted in the article. They build gearboxes (among many other things) that are used by many different vehicle manufacturers. These gearboxes have an "API" for lack of a better term, such that while many cars have the same mechanical part, the individual manufacturer is able to tune line pressures, shift maps, shift speeds, rev matching etc to their own particular liking. A sports coupe and a luxury sedan may use the same transmission, but they are tuned completely differently for different audiences. This has the undeniable advantage of economies of scale - not every manufacturer has to build their own bespoke transmissions. It also has advantages for the end user who likes modding - I can flash custom parameters to my ZF transmission without any difficulty as the API is a known quantity and software is readily available.
The same is true of ignition systems, braking systems, and many other subsystems. It just does not makes sense for everyone to make their own, they just buy one and customize the software for their particular application. Parts are more available, service knowledge is more available, and it is cheaper in the end. I'll take this approach over the in house mysterious black box approach any day.
Re: (Score:3)
The first company to break with this philosophy was Tesla, which has from the beginning had a strong philosophy of in-house software design
Ford has been doing their own PCMs for decades, since before Tesla was even thought of. My first real tech job was with a company which designed silicon for other companies... including Ford.
Re: (Score:2)
The level of technology depends on what feature set you expect.
IMO reliability peaked in Japanese cars of the very late 1980s through the 1990s. But if you want no computer you should get a diesel Mercedes. Don't buy a cheap one, though. They're not worth it.
Bikes without code? (Score:2)
Not too long ago I bought a new road bike (currently off the road because of a vandal) and almost the first thing I asked was "does it have anti-lock braking on the front wheel?"
I don't know if you've ever ridden a bike on the road in the wet, but an emergency stop on a bike used to be a gamble - squeeze too hard on that brake lever and you're _off_. These days I have mixed feelings about the days I used to get around this city on a sports bike _without_ anti-lock braking. They were special times. I was fo
Re:Cars without code? (Score:5, Insightful)
What was the last car without this nonsense? Makes me want to buy a classic car or motorcycle, just for the simplicity.
Those were the days! When people would stop and open a bottle of champagne if the car made it to a hundred thousand miles, you had to change the oil every 2K miles, the points and plugs several times a year, the distributor every other, and I remember tires lasting around 10K miles.
Part of what makes modern cars last longer is all of the computerized geegaws in them. My car gives me readout of exact water, oil, and transmission temperature Individual tire pressure. Integrates with my phone and map, has voice control of the maps and calling Backup camera with position guides, and a whole bunch of other stuff I'm probably forgetting.
Now you do have some options. Some folks restore cars, so if you want to go mid 60's you can get some thirsty vehicles, but nicely restored. If seat belts are annoyind and you don't want them, Go to the early 60's. The mid to late 50's has a lot of iron that isn't encumbered by safety at all, and there is a divide around anything before 1955 that is pretty pure vehicles - no safety, no automatic chokes, a lot of standard transmissions. And no computers.
Re: (Score:2)
The last vehicle I had without this crap was a Canadian '86 720 Nissan truck. Had an electric choke and an electric solenoid to turn off the gas when decelerating or turning off the engine. Choke was fixed with a cable and the other thing wasn't needed though it did lower emissions when decelerating.
Before that I had a '84 diesel of the same model that given a hill to start on, didn't need any electricity if you didn't mind no lights, the fuel pump shutoff had been replaced with a cable.
Re: (Score:3)
What was the last car without this nonsense? Makes me want to buy a classic car or motorcycle, just for the simplicity.
Sort cars by death rate of drivers, and pick the highest. You'll find the car without code there.
Re:Cars without code? (Score:5, Informative)
Hahaha, actually fuel injection was invented in the late 19th century, and as for gasoline engine automobiles Chrysler produced those starting in 1958. You don't need one line of code to have electronic fuel injection.
I see opportunity for cars without all this unnecessary digital digital mental retardation, it's getting way out of hand. 100 million possible points of failure is more like the truth.
Re: (Score:2)
Re: (Score:2)
I could also see "maker" group forming that would develop procedures to rip all that crap out of an engine compartment of certain vehicles, and transform into a pure analog beast. For some crazy (and bad) Rube Goldberg engine designs it's not possible, they require a computer but that's another rant.
Re:Cars without code? (Score:5, Insightful)
I could also see "maker" group forming that would develop procedures to rip all that crap out of an engine compartment of certain vehicles, and transform into a pure analog beast. For some crazy (and bad) Rube Goldberg engine designs it's not possible, they require a computer but that's another rant.
I think you're throwing the baby out with the bath water. I also like the old days when you could open the hood of a car and actually see the ground underneath. But computerization has brought a LOT to the table in the way of fuel efficiency, performance, consistency, and even - yes - reliability. That said, I agree that ECU's have become overly complicated.
Design-wise, there are three major things automakers are doing today that I have a problem with. The first is trying to protect us from ourselves. Mostly, they need to just fuck off with that shit. Teach people good solid driving skills, test them rigorously before you hand them a licence, and enforce the important laws seriously and consistently. We shouldn't be depending on automation to replace skillful, mindful driving and enforcement thereof.
The second thing they need to get over is tying infotainment and critical systems to the same comms bus. These should be totally separate. If infotainment needs drivetrain info, the communication should be one way only - drivetrain ---> infotainment. No handshaking and no acks - it shouldn't be possible for infotainment to even query the drivetrain or critical safety systems, nor anything that affects their operation.
The third thing is OTA updates of engines and safety systems. That should be utterly against the law - the capability shouldn't even exist in any production car. Violation of that law should carry a minimum $100 million fine for the first offence, escalating thereafter.
Re: (Score:3)
Mostly, they need to just fuck off with that shit. Teach people good solid driving skills, test them rigorously before you hand them a licence, and enforce the important laws seriously and consistently. We shouldn't be depending on automation to replace skillful, mindful driving and enforcement thereof.
Just pure BS. The most important discovery of the last millenium in the safety science is that humans can't be trusted. That's why safety improvements pretty much never come from "education".
The second thing they need to get over is tying infotainment and critical systems to the same comms bus. These should be totally separate.
They already are. Modern cars often have 3 or 4 CAN buses: for the engine and critical components, the general infotaiment and car management and usually a separate always-powered CAN bus for stuff like radio door openers.
The third thing is OTA updates of engines and safety systems. That should be utterly against the law - the capability shouldn't even exist in any production car.
So you want people to die because the manufacturer can't push a critical fix?
Re:Cars without code? (Score:5, Insightful)
Teach people good solid driving skills, test them rigorously before you hand them a licence, and enforce the important laws seriously and consistently. We shouldn't be depending on automation to replace skillful, mindful driving and enforcement thereof.
You can't teach people perfection. You can't teach people to not be human. Automation exists because the industry is clever enough to realise that regardless if you're selling cars in a some backwater part of Africa where licenses are unheard of, or in Germany where the driving license comes with many hours of professional driver training and many more hours of theory, in both cases the automation has dramatically improved safety.
This applies equally to flying planes, operating major hazard facilities and any other activity that humans undertake. It's an absolute fantasy to think that you're in any way better than the software that can control your car. We've seen that time and time again, we've pitched professional drivers against amateurs in cars where the systems can be toggled only for the professional drivers to get schooled as to how irrelevant their skills are in comparison to modern safety systems.
Yet the myth persists. You'll always find someone who thinks they can out stop an ABS system telling people to pump their breaks or use some fancy technique which they have zero chance of ever doing in an emergency. No. YOU NEED PROTECTION FROM YOURSELF. You like I are nothing more than a stupid crappy fallible human.
Re: (Score:2)
I could also see "maker" group forming that would develop procedures to rip all that crap out of an engine compartment of certain vehicles, and transform into a pure analog beast. For some crazy (and bad) Rube Goldberg engine designs it's not possible, they require a computer but that's another rant.
Gonna have trouble with emissions.
If modern cars are offensive, best to buy something from the 50's, pre computer, pre smog, pre safety - just pure riding pleasure unencumbered by advancement.
Re: (Score:2)
And then don't crash it, because that car is designed to have you scraped out of it in chunks while the car survives.
Re: (Score:2)
Re: (Score:3)
The mechanical fuel injection (Rochester Ramjet) on the Corvette was even worse than early analog fuel injection systems in terms of 2 times out 10, your car just refusing to even start.
Re:Cars without code? (Score:5, Insightful)
and as for gasoline engine automobiles Chrysler produced those starting in 1958. You don't need one line of code to have electronic fuel injection.
The Bendix Electrojector and subsequent Bosch D-Jetronic truly sucked ass and are immediately replaced by a carburetor by anyone unfortunate enough to own one of those systems (unless they're trying to keep it stock)
The later K-Jetronic is similarly almost immediately removed if you can.
Those early systems were necessary for the evolution of the system that anyone with 2 brain cells knew would eventually be far better than a carburetor in terms of stoichometric burning and overall control of the fuel mix in general, but they absolutely sucked ass until the full closed-loop Bosch Motronic (microcontroller based) systems came out in the 80s. (79, technically)
In summary- yes, you do need lines of code to run fuel injection worth a damn, and every system prior to that is nearly universally looked at in horror for the problems they had.
Re: (Score:2)
My Rabbit had mechanical fuel injection. I kept that car for 34 years and can't remember any problem with the fuel injection system except leaks at the injectors.
Five transistors in the entire car, all of them in the radio.
Re: (Score:2)
The Corvette's system used vacuum logic, and suffered from unreliable choking ability leading to the car just not starting frequently.
As for the Jetronics, those fucking things would just lose their "mind" and run obscenely rich or obscenely lean for no discernable reason that could only be fix
Re: (Score:2)
You were lucky you never had problems, or you just didn't care about them.
If you had a single gummed up valve or vane anywhere in the air or fuel control feedbacks (which happened frequently) it would just not run (or you'd get the characteristic backfires)
You could fix them easily enough, but you'd have to take the entire thing apart, because there was absolutely no way to discretely identify the problem
Re: (Score:2)
Perhaps it was a diesel? I had a '84 Nissan truck, diesel, pure mechanical fuel injection, and the only problem I ever had with the fuel system was needing to replace some fuel line.
Re: (Score:3)
The Bendix Electrojector and subsequent Bosch D-Jetronic truly sucked ass and are immediately replaced by a carburetor by anyone unfortunate enough to own one of those systems (unless they're trying to keep it stock)
Which is quite ironic given that Bosch is the name in fully mechanical diesel fuel injection. The so-called "P pump" (P7100) as well as the M and MW pumps are rock solid, the injectors have long lifespans when used with quality fuel, and the glow systems are highly reliable when not left to soak in diesel fuel from the admittedly flaky return line systems :)
Re: (Score:2)
It was a good system when it worked.
The problem was never the pumps or the injectors, but the complicated system of control plungers and vanes that were prone to getting sticky and required a complete teardown to fix, and could happen pretty frequently.
Of course, the system would also generally run, even when it was performing poorly, so most people just didn't care.
You still see K-Tronic mercs rolling down the road smelling like shit, and the people driving
Re: (Score:2)
Re: (Score:2)
100 million possible points of failure is more like the truth.
In a way, yes, but the upside is that there's no "wear and tear" on software. The software will undoubtedly ship with flaws, but with over-the-air updates most of those flaws can be repaired as soon as they are detected and understood, and eventually (decades from now) the codebases will be mature enough that very few bugs remain.
In an analog hardware-only design, OTOH, even the most skilled engineering will result in moving mechanical parts that mesh together perfectly for a while, but then start to lose
Re: (Score:2)
The first fuel-injected cars are probably when this started: https://www.chipsetc.com/compu... [chipsetc.com]
Mid 60's Corvette had a fuel injection option. It was a cool if fussy ride. But no computer.
That explains the chip shortage (Score:4, Insightful)
Got word today that orders we placed at the beginning of March (literally, the first few days) are now scheduled for arrival in November. No, that is not a mistype. Eight months to get an order of PCs. And that's only one of the orders placed within a day or two of each other. No word on when those other orders might come in.
All because people are too lazy to learn how to properly drive a car and car manufacturers see an opportunity to soak the stupid. Having driven a vehicle with adaptive cruise control, it sucks balls. Continually slowing down then speeding up could give some people sea sickness. Automatic emergency braking? The last thing I want is some jacked up programmer determining when to brake my car. Lane control guidance? Sure, let's have a piece of software yank the wheel in a direction the driver isn't expecting or prepared for, especially when they are nowhere near the center lane line or the outside line.
Just remember, the more complicated you make something, the more expensive it is to repair. When those chips start going out or an "update" borks something, you get to pay for it.
Re: (Score:2)
My Mahindra (aka curry muncher) has simple, almost analogue software,
manual gearbox, no bluetooth, gps, cameras, software centre display, etc.
you really drive this car, old school
Her Mazda BT50 has all the terrible software "smarts"
I hate the steering wheel tugging, the unexplained braking, alarms,
distractions, poor 360 deg. vision, locking out controls while driving,
the car knows better than the operator, but some of the software "features"
are just dangerous.
I recently had to scrap a Renault van when it co
Re: (Score:2)
I don't know the details but I heard a rumor that some French auto makers were getting some bad product from some of their German suppliers.
That sort of thing might explain your Renault problem.
Re: (Score:2, Offtopic)
This chip shortage was not caused by automakers. By and large, the surge in new demand for fab capacity was predomaintly ASICs and GPUs for crypto mining.
Re:That explains the chip shortage (Score:5, Informative)
Why There are Now So Many Shortages (It's Not COVID) [youtu.be]
So no you can't blame it all on cryptomining any more than one can blame it on that other boogieman, the cloud.
Re: (Score:3)
There are shortages because people insist on using videos for things that can be described easily in text, and video requires a lot more CPU
Re:That explains the chip shortage (Score:4, Insightful)
Continually slowing down then speeding up could give some people sea sickness.
I used to get that with regular human drivers with no adaptive cruise control.
Automatic emergency braking? The last thing I want is some jacked up programmer determining when to brake my car.
I wouldn't want some overconfident arsehole, who believes themselves to be one of the few who learnt how to "properly drive a car", in full control of their "braking" either.
Re: (Score:3)
I wouldn't want some overconfident arsehole, who believes themselves to be one of the few who learnt how to "properly drive a car", in full control of their "braking" either.
Exactly. The automatic braking isn't there so you don't have to pay attention. It's there in case you fail to do so. It's almost like people forget that humans are highly fallible.
Re: (Score:2)
Re: (Score:3)
The amount of accidents was never in acceptable territory. Not even once. To back up that claim, I offer:
- mandatory seat belt laws
- mandatory side mirrors
- mandatory anti-lock braking systems
- mandatory blind spot monitoring / alerting
- mandatory reversing cameras
- mandatory traction control / stability control systems
- mandatory crumple zones
- mandatory active head restraints
- mandatory tire pressure monitoring
- mandatory air bags
etc.
If they were "acceptable levels" then why do we keep increasing the am
Re: (Score:3)
Just remember, the more complicated you make something, the more expensive it is to repair. When those chips start going out or an "update" borks something, you get to pay for it.
Remember? Why do you think they're doing this? It's by design.
This is what happens when you sit back and allow Greed to turn into Relentless Greed. Standard maintenance and repair wasn't quite profitable enough for a stealership. Now, they include all manner of bells and whistles as standard features on most cars, driving up both the profit on the front end (car sale), as well as on the back end (tripling maintenance and repair costs).
And then there's the stupid shit. Like a $5K bumper attached to ada
Re: (Score:2)
Just remember, the more complicated you make something, the more expensive it is to repair. When those chips start going out or an "update" borks something, you get to pay for it.
That's just not true. Car maintenance is now cheaper (in inflation-adjusted dollars) than in the blessed pre-computer 70-s.
E.g. look at your central door lock control. In 70-s that would have required multiple individual wires from the door locks to buttons on the dashboard. And the buttons themselves were likely latching switches. These days door locks are wired into the CAN bus and don't need dedicated wires. Buttons are cheap momentary-contact switches that are also wired into CAN bus. Breakdowns can b
Re: (Score:3)
Just remember, the more complicated you make something, the more expensive it is to repair..
Most of this complexity is massive improvements in safety features. Avoiding collisions is a great way to reduce repair costs.
Also, the people inside the car are far harder and more expensive to fix after a collision. Shifting the damage from the people to the car is a huge win.
Re: (Score:3)
Having driven a vehicle with adaptive cruise control, it sucks balls. Continually slowing down then speeding up could give some people sea sickness.
Computer analogy (because we're talking about cars): You used the Windows Me of adaptive cruise controls and decided based on that Linux isn't reliable or stable enough for you.
Automatic emergency braking? The last thing I want is some jacked up programmer determining when to brake my car.
Which was the last thing he said as he inattentively plowed into the back of the car infront of him.
You're an idiot. Worse, you're a dangerous idiot. You deny the basic statics which point to you as a human being more unreliable than the electronics we have used to dramatically lower the death toll. And you like 95% of people ever as
no non dealer repair for you! (Score:5, Insightful)
no non dealer repair for you!
That line count sounds exaggerated (Score:5, Interesting)
Re: That line count sounds exaggerated (Score:3)
I have a hard time believing that my F-150 has 150k lines of code. If it does, then they must be largely NOPs or as you say the code count is including the RTOS.
Re: (Score:3)
I say if they're charging $70K for a pickup truck, they *better* be including at least 100 million lines of code.
Otherwise, it would be a real ripoff.
Re: (Score:2)
I'm not sure where the fuck that number comes from, but in our many, many discussions about ECU firmware, he never once mentioned something that insane.
I suspect someone decided to count "Lines of assembly" or something equally fucking stupid.
How are they counting ... (Score:5, Insightful)
Now the old joke about your car (Score:2)
Showmanship. (Score:2)
Ask the military on how to write in ADA.
Re: (Score:2)
You can ask Toyota how the verifiable SPARK subset of Ada is working for them: https://www.adacore.com/custom... [adacore.com]
Re: (Score:2)
Why, oh why... (Score:2)
Re:Why, oh why... (Score:5, Funny)
That '78 Thunderbird may be one of the most grotesque and excessive cars ever made. The needless pop-up headlights and the 2 doors so long that you need to park next to an empty space just to get in and out. A boxy "personal luxury car" body on an LTD frame. Velour seats. The 400 cubic inch Cleveland engine (or the slightly smaller Windsor) got about 3 miles to the gallon. It was all so tasteless and decadent. It is a car that positively calls out for a man who wears Brut cologne and leisure suits.
I would totally drive that car.
Re: (Score:2)
I had to go look for this behemoth [google.com]. Similar to the Lincoln Mark V. [google.com] (apparently one went for 18K USD last year [hemmings.com], per Hemmings).
Re: (Score:3)
Probably because it was a pile of shit in comparison to modern vehicles. You've just forgotten how many headaches it gave you.
But you're right, at least you could get your hands dirty and fix it yourself when it was running rich instead of having to figure out what sensors cause what feedbacks in an EFI system.
Buy another vintage ride. (Score:2)
Still cheaper than new and there are plenty of parts. Be willing to travel for a deal and collect donors if ya have room. (I live in a rural location for "freedom to wrench" reasons.)
Forums are a great source of info and remember fuel mileage is only PART of operation and maintenance costs. Wrenching is so damn much fun why deny yourself? Earlier is better from a "classic vehicle" exemption POV and American cars ain't shit compared to the old British motorcycles I'm masochistic enough to collect.
When vehicl
I got a sixty-nine Chevy with a three-niner-six... (Score:2)
the little-known successor to the 386
Insert 50 dollars to continue... (Score:2)
We are slowly getting closer and closer to more companies pushing subscription services for vehicles. I know a few have tried and might be doing it to a small degree, but give it 5-10 more years and you're going to see more players trying to sell you a subscription that allows you to drive your car outside of business hours or outside a restricted area.
"Please accept the billing of $1999.99 for travel to California"
Re: (Score:3)
Like trying to buy a "dumb" TV (Score:2)
I don't doubt that there are genuine investments made in safety and quality mechanisms. These would be good for the end user, mostly, and might be worth paying for. But it's also very much in the interests of manufacturers to continue to "innovate" and ultimately complicate their products. It's a key part of their marketing and helps justify high prices. Putting out a dumb and simple product, be it a car or a TV, would eventually lead to much lower prices and selling a cheap (albeit entirely adequate) produ
Re: (Score:2)
...Putting out a dumb and simple product, be it a car or a TV, would eventually lead to much lower prices and selling a cheap (albeit entirely adequate) product is not good for business.
Turning every car into a $30,000 investment that not everyone can afford, is not good for the economy.
Unlike HDTVs, cars are a rather necessary component of modern survival for the overwhelming majority.
...in a western market where consumers have shown a willingness to pay far far more.
No one ever said they were smart about it. Rampant narcissism doesn't mean you simply try and keep up with the Joneses. It means you target and destroy the competition. Debt, has become a fashion statement.
Software requires hardware (Score:2)
I'm holding off a decade (there is no NEED to buy a new car for most private owners and buying early is like buying a 1920s auto whose designs were swiftly superseded) or so for the shakeout.
We don't know what will last as we do with older systems and I don't spend money due to moral panics. I let others beta test and prove what is supportable (the most-produced, most widely popular systems proven over time and effectively supported over long service lives).
I want everyone excited by new toys to buy them. T
Re: (Score:2)
A new vehicle should be good for twenty years outside the Rust Belt.
Imagine if smartphone hardware was as rugged as a G-shock watch, and the batteries lasted at least a decade or more.
You would still have a major problem with trying to maximize product life.
Premature obsolescence, will feed Greed no matter what you want.
If the glorious Casios were built (Score:2)
with superior materials they'd last even longer. Casio are so brilliant because they pull it off on a budget so low that manufacturing cost must be near-trivial. Imagine what Casio could do if they made the same watch at a say a ten dollar cost to produce. The damn things would last decades easily and battery replacement is already easy.
I still have my father's old Rolex but consider Casio a far greater engineering accomplishment.
BTW here's Casio battery swap info: (Score:2)
I forgot to include how to swap (this is or was a tech site) your own, and this forum is a goldmine of watch info and there are more worthwhile Casio posts:
https://www.watchuseek.com/thr... [watchuseek.com]
And what do we know about the quality? (Score:2)
How about ten thousand global variables?
https://news.ycombinator.com/i... [ycombinator.com]
Then there's the security problems. I met someone at a conference who kept finding exploits that would let people steal cars. He said the car companies completely ignored him. I suggested he contact the insurance companies.
On the other hand, I don't think I could live with the standard of safety features on classic cars. I have a friend who had a blowout at 70 mph IIRC in poor conditions and he thinks the stability control is the reaso
Re: (Score:2)
I had a blowout at 70 mph, and I'm pretty sure AWD is the reason it was just a "pull over and stop while making a lot of noise" moment rather than a loss of control.
Lines of code? (Score:2)
Re: (Score:2)
Dont we have any better way to measure the amount of code?
Yes:
(number_of_gotos / number_of_comments) == level_of_awesomeness
Just give me a basic car (Score:2)
I really dont want all this tech already in my car. I don;t want nav (I can get that on my phone), I don;t want self-driving, I don't even want an LCD screen (that will ultimately burn out leaving your car unusable, and cost thousands if you can still even get one when it inevitably craps out).
I just want a minimalist car with only the basics needed to drive around and blow hot/cold air into the cabin, and a standard stereo slot so I can buy/fit an aftermarket stereo, so that I can keep up with emer
100M+ lines of code? (Score:2)
Simple rule to remember (Score:3)
I don't believe 100 ECUs per car (Score:3)
My 2008 diesel Jaguar (OK, old, but with lots of electronics,) has about 20 ECUs; about the only thing it's lacking over newer cars smartphone integration. That 100 looks to me to be including electronic "modules" such as speed sensors, wiper motor switches etc, as well as ECUs.
The referenced IEEE document is from 2009, so it should be comparable. If they've overcounted the ECUs by a factor of 5, I guess that will be true for the LoC.
Keep It Simple Stupid! (Score:4, Insightful)
Ugh that acronym means something else damn it (Score:4, Funny)
ECU already means engine control unit. It's what we called the computer that controlled the engine before 1996, when we went OBD-II and started calling it a PCM. (PCM is the name literally mandated in the OBD-II specification; some were using it already, most weren't.)
Using it to mean "electronic control unit" is stupid and misleading, because it already means something else.
Re: LoC is a bad metric, as always (Score:5, Informative)
Having worked in the vehicle industry with development the amount of code hardly surprises me. And not all code is C/C++ but instead Simulink and sometimes Java. There may be different code languages too.
Of all alternatives Simulink is worst because it's hard to refactor and it generates C code which might yield different output depending on version/patch level of the tools.
Then the software development is distributed over the world so people that don't even know about each other have to make it work.
The operational principles are also different - all code is always running, even if there's no change in the input. So to read a switch it's polled every 10ms. The result is that the ECUs uses more electricity than necessary.
In later years vehicles also uses FPGAs that interacts with the processors. Just trust that the FPGa grid works properly or you'd end up popping the trunk lid when you turn left while braking and pushing the recirc button.
Re: (Score:3)
Re: (Score:3)
I always wondered about the extra wear on the starter and battery (which, afterward, needs extra fuel to recharge!). Is replacing those items "greener" than just letting the running engine idle?
The wear is minimal because it's easy to start a hot engine, and the surfaces in the engine are still well-coated with synthetic oil (which is specified for pretty much everything now, especially vehicles with auto start-stop.)
Not to mention, all the extra costs associated with the technology (creation, upkeep) to do such a thing.
Those costs are roughly dick. Modern vehicles already control the starting process, especially the ones with a start button.
The best auto start-stop solution is a mild hybrid, which replaces the starter and alternator with a starter-generator, and uses a relatively small lithium batte