Slashdot Mirror
VW Admits Audi Automatic Transmission Software Can Change Test Behavior (cnet.com)
In response to a report via Bild am Sonntag last week, which found a new type of defeat device hidden inside an Audi automatic transmission, Volkswagen finally came around to admitting the findings. "Adaptive shift programs can lead to incorrect and non-reproducible results" in emissions tests, VW told Reuters on Sunday. CNET reports: Software in the AL 551 automatic transmission may detect testing conditions and shift in a way that minimizes emissions, only to act "normally" out on the road. Much like Dieselgate's defeat device, that leads to higher-than-imagined pollution, which could be in excess of legal limits. Audi's AL 551 can be found in both gas and diesel vehicles, including the A6, A8 and Q5. Volkswagen isn't going full mea culpa here, though. The automaker also told Reuters that its adaptive transmission software is meant to change shift points in order to improve on-road performance. Many automatic transmissions these days learn from driver input and tailor shifting to match a driver's style, which leads to a smoother drive. VW Group did not immediately return a request for comment.
Hey, the Hay Sniffer is a legitimate hack. This is the software routine that "sniffs the hay" to determine if you are out on a country road and not driving one of the Federal Cycles.
My criterion is if you drive a Federal Cycle for real out on a highway, a test track, or a high school parking lot, it should give the same control coefficients as on the chassis rollers in Ann Arbor, Michigan. None of this "oh, only the back wheels are turning, I must be in Ann Arbor."
But if it only gives Federal Cycle performance if you actually drive that way, good. If it gives you different performance for driving "off cycle", so much better.
For $deity$ sake. Is this the final word ?
Or tomorrow we will find that in test conditions it transform also in a unicycle ?
This just keeps on getting better and better. VW Group have simply not owned up to the depth of their cheating and been forthright with their cooperation.
Our regulators should slap increasing penalties on each successive cheat they find, to penalize for the hiding of evidence over and above the violation itself.
"can lead to incorrect and non-reproducible results". I'd be bloody rich. I'd have just said: "Yeah, you got us again. How much you want this time?"
Hi! I make Firefox Plug-ins. Check 'em out @ https://addons.mozilla.org/en-US/firefox/addon/youtube-mp3-podcaster/
I have a Landcruiser that has 'adaptive' shifting. A button I press in for power and out for economy. Want to hear something worse? Everything else I own is a stick. So I shift when I damned well please.
Have gnu, will travel.
No nothings.
This is BS. There are so many variables in emission testing that almost anything will affect the results. I's sure altitude, humidity, gas octane, maybe even oil could affect things to name a few. Give it up!!
The only question I want answered is "did it pass the test as written by the government?". If yes, what's the problem. If you don't like the results, fix the test.
I hate to say it, but physics is a bitch, and politicians are full of shit. It's not possible to build a vehicle that meets current emissions standards without cheating or pricing the middle class out of vehicle ownership. It's just not possible. This is not VW cheating; it's the politicians lying.
The sooner we as a society can move to electric, the better.
"Nine times out of ten, starting a fire is not the best way to solve the problem." - my wife
but there are no such thing as defaults in a car that even gets driven to the test place. how hard you jam down on the pedal affects the "sportiness" and thus shifting points, all kinds of things affect it.
the problem is that they do the tests like this: stick it on a dyno and run a pre made program, without even fucking moving.
they should just make a tester small enough to fit in the boot or passenger area, and drive around a test track - vw's "cheating" would have had to be of different kind in that case.
also non-aerodynamic SUV's would be paying their fair share of the taxes.
now here's a test to test if this was malicious from vw or not: drive around using the pedal same way the test does. having an adaptive gearbox isn't cheating the emissions any more than having a manual and driving really slow is.
and why the tests cannot be changed in most countries easily nowadays: THEY AFFECT CAR VALUES DIRECTLY, ever since they started to move to co2 based taxing based on the test - changing it to an actual test could disrupt car prices 30-50% UP AND DOWN.
world was created 5 seconds before this post as it is.
To see something like your post on Slashdot, of all places, is some serious shit. I thought Slashdot is where intelligent people that are aware of the problems are. Apparently not. It's obvious that if the software in question was Free Software, this could not have happened. And you reply with some ad hominem (I think?). Sigh, Slashdot was good once.
So much for the famous german quality. I guess they entered the realm of globalization, and got really prolific at it (importing workers by the million, exploiting existing trademarks but putting inside imported stuff, cheating at tests, etc.).
How much of the behavior of the gearbox is from unique software that is only found in Audi products. How much is software ZF the gearbox manufacturer uses in all or most of their 8HP55 units. ZF supplies just a few Auto manufacturers with their 8 speed auto.
List of ZF 8HP variants
Name Max. gasoline torque (Nm) Max. diesel torque (Nm) Application examples
8HP30 300 300 -
8HP45 (a separate 845RE is by Chrysler)[9] 450 500 BMW 1 Series (F20), BMW 3 Series (F30), BMW 5 Series (F10), BMW X4 (F26), Chrysler 300 V6, Dodge Charger V6, RAM 1500 V6 (2013- )
8HP50 (2nd Generation)[10] 500 500 Alfa Romeo Giulia, BMW 1 Series (F20 LCI), BMW 3 Series (F30 LCI), BMW 5 Series (F10)
8HP55 550 550 Audi Q5, Audi Q7, Audi A8 and A8L (D4)
8HP70 700 700 BMW 7 Series (F01), BMW X5 (E70), Dodge Durango V8, Iveco Daily, Jeep Grand Cherokee V8 & diesel, Ram 1500 V6, V8, & diesel, Range Rover Sport (2012), Rolls-Royce Phantom, Maserati Quattroporte (2013–), Dodge Charger V8 (2015-), Dodge Challenger V8 (2015-), Aston Martin Vanquish (2015), Aston Martin Rapide (2015), Aston Martin DB9 V12 5.9L (2012-)[11]
8HP75 (2nd Generation) 750 750 BMW 7 Series (G11/12),[12] BMW X5 (F15)[13]
8HP90 900 1000 Rolls-Royce Ghost,[14] Rolls-Royce Wraith (2013), Bentley Mulsanne, Dodge Challenger SRT Hellcat, Dodge Charger SRT Hellcat, Bentley Bentayga, Audi RS 6, Audi SQ7
With NOx... forget companies like M$ who monopolise their market, this is selling something that causes lung cancer to passers by. Someone fucking end this company.
Please go back to /g/.
Change is certain; progress is not obligatory.
Given the latest revelations, I doubt there a product they make that we can definitely say is actually compliant.
At this point, it's a plateful of lies smothered in bullshit sauce.
The only penalty regulators should be considering at this point is shutting them down.
I suspect VW is not alone in trying to skirt emissions. Now a couple reports say some Dodge truck owners are also claiming emission issues. As a person who has worked on vehicles all my life, I am not surprised that manufactures feel overwhelmed by the emission dilemma which can choke a vehicles performance in order to meet the standards. This ends up being a problem for a vehicle manufacture trying to satisfy a customer more concerned about performance and drivability than emissions. I can almost guarantee the fine line of meeting emissions is only obtainable in a predictable test and would many times fail a real world road test.
The games that VW and other manufacturers play with emissions did not kill anyone the way that a wheel bearing or steering rack that suddenly fails at 60mph would. Were heavy metals released into the atmosphere in a quantity that would acutely affect someone's health? Were toxic chemicals leaked into ground water, contaminating it?
And what would happen to this company if it was shut down? Do you think that it has more value as a single unit, or more value as a group of disjoint businesses? Do the buyers have the right to keep it together? Most importantly, what about the vast number of VW employees who acted with high ethical standards and completely within the law? Do they all of a sudden get thrown out on the street? What will happen to the economies of the countries that host their labor?
I'm all for jailing and confiscating personal assets of the decision makers and active participants in this, but it needs to be proportional to the crime. What you're proposing is ludicrous when nobody has died, been seriously hurt or become ill, particularly when it is based on a number from some government agency that someone exceeded.
Let's see... say 10,000 cars sold in the US per month since Jan 2010, that's 720,000 cars, say average selling price $20,000... VW committed $14.4 billion of fraud selling Audi alone. And that's just the compensatory damages, punitive damages and even criminal penalties are where boot meets neck.
Say it ain't so!
Not having access to the source of the engine management software and related, and the ability to verify that the software in cars does derive from said source, is a major problem. Software is potentially too unpredictable to test in a black box way, and cars are dangerous things. Independent auditability should be a must for software related to how a car drives.
DRAIN THE SWAMP!!!!
From the description, this doesn't sound like cheating. It simply sounds like the transmission shifting algorithm can vary shift points, which in turn can affect emissions. There's nothing surprising or revelatory about that. The real problem seems to be that the EPA is using a static test for a dynamic system.
The point was that the software is adaptive. If you stomp throttle (like God intended) the slushbox gives you RPM, if you pussy foot (like when on an emissions dyno) to get a consistent RPM, the transmission will shift quickly to top gear. Intended behavior for a transmission being called cheating.
Old school slushboxes did basically the same thing with analog computers. Engine manifold vacuum, modulator and springs in the transmission valve body.
Digital computers just adjust shift points based on past driving. I bet the computer doesn't produce 'smoother shifts' when someone like me is driving it, I bet it shifts more like it had an old school shift kit...I bet that would annoy the slushbox owner, once I get back into my car with a real transmission.
They are basically bitching that their test doesn't reproduce real world driving. No full throttle testing to yellow line.
John McAfee 'It was like that time I hired that Bangkok prostitute; to do my taxes, while I fucked my accountant'
My 2001 Camaro SS has a physical lock-out that forces you to shift from first to fourth if you're driving like an EPA tester, a "feature" which is actually rather difficult and unsafe to demonstrate on the road.
My Nissan Leaf does not cheat on emissions tests because it has no exhaust. It also does not have a transmission, so it can't cheat that way either. The closest that it is capable of cheating is the Guess-O-Meter, which determines the remaining drivable mileage based on various factors.
The point is that modern cars have computers in them running proprietary software which control how the car behaves. Implementing the same limits with mechanical apparatus means exposing how the apparatus works and allowing the car owner to remove or adapt the apparatus. Proprietary software hides the rules and makes it much harder for the car's owner to remove or adapt how the software works. Apparently a variety of car manufacturers use this secrecy to deceive consumers into buying a car that didn't behave the same in testing as it does in regular use. The consequences of this are vast and hardly limited to cars. But the only real solution is the same: pass on the code to the good being sold under a free software license right along side selling the good so the owner of the good can truly make their object their own.
"Bitching that their test doesn't reproduce real world driving" is very much the problem here because the same thing happened in environmental tests with a very large variety of makes and models running software designed to cheat testing. It hardly matters whether the feature that exposes the problem is compliance with emissions regulations, getting the RPMs one expects out of a car, or anything else because the underlying issue is controlling the user through proprietary software and therefore one has to understand the inherent untrustworthiness of proprietary software as the root of the problem.
If you see the commonality between this story and so many other stories on /. it's because you understand that /. points readers to a lot of stories where proprietary software is to blame. Every DRM story, every story where the good the owner purchased isn't behaving reasonably comes down to proprietary software isn't giving the owner full control over the device they own. Anyone who owns anything running on proprietary software has good reason to be concerned about this. Everyone should use the presence of proprietary software in a device as a reason to not buy that device. The only way to fix that problem is free software. As I said in another thread, software freedom is the only thing that will keep proprietors from taking advantage of computer users because when the proprietors don't know who is inspecting the code, improving the code, or distributing improved versions they know they can be caught.
Digital Citizen
US is a little fuzzier. The laws are written to take intent into consideration. In other words, if you follow the letter of the law and violate the spirit you get nailed. It's banking, which hurts people that matter though, so take that as you will.
Hi! I make Firefox Plug-ins. Check 'em out @ https://addons.mozilla.org/en-US/firefox/addon/youtube-mp3-podcaster/
My BMW M3 does much the same thing with automatic mode in it's SMG gearbox which is basically a manual double clutch transmission with the option of allowing a computer to shift for you. It will change and adapt over time the more you drive it. Gearheads have known about this for over a decade now.
Whatever they're paying these people to inspect these cars, they're paying them too much if they don't know about stuff like this.
BeauHD. Worst editor since kdawson.
On VCL threadsafety, your stupid use of Application.ProcessMessages (vs. progressbar .update or .refresh which VCL controls have) & even Sleep API direct vs. Application.ProcessMessages (+ even DOING a looped load, lol, when IT IS NOT NECESSARY!) to cede cpu (slowing your UNNEEDED loops too, lol, & you talk speed? Please)!
* I am about ACCURACY in work with data - YOU, clearly, ARE NOT (accuracy over speed? I'll take the former, EVERY time).
You're also INCREDIBLY inefficient!
APK
P.S.=> It's all here in black & white, see subject https://apple.slashdot.org/com... chump... & when YOU DO A BETTER PROGRAM OF THIS NATURE (imitation IS the sincerest form of flattery & you would be imitating me)? Then, talk (you've already TALKED OUT YOUR ASS taking shots @ me & I shot you down for it easily)... apk
On VCL threadsafety, your stupid use of Application.ProcessMessages (vs. progressbar .update or .refresh which VCL controls have) & even Sleep API direct vs. Application.ProcessMessages (+ even DOING a looped load, lol, when IT IS NOT NECESSARY!) to cede cpu (slowing your UNNEEDED loops too, lol, & you talk speed? Please)!
* I am about ACCURACY in work with data - YOU, clearly, ARE NOT (accuracy over speed? I'll take the former, EVERY time).
You're also INCREDIBLY inefficient!
APK
P.S.=> It's all here in black & white, see subject https://apple.slashdot.org/com... chump... & when YOU DO A BETTER PROGRAM OF THIS NATURE (imitation IS the sincerest form of flattery & you would be imitating me)? Then, talk (you've already TALKED OUT YOUR ASS taking shots @ me & I shot you down for it easily)... apk
On VCL threadsafety, your stupid use of Application.ProcessMessages (vs. progressbar .update or .refresh which VCL controls have) & even Sleep API direct vs. Application.ProcessMessages (+ even DOING a looped load, lol, when IT IS NOT NECESSARY!) to cede cpu (slowing your UNNEEDED loops too, lol, & you talk speed? Please)!
* I am about ACCURACY in work with data - YOU, clearly, ARE NOT (accuracy over speed? I'll take the former, EVERY time).
You're also INCREDIBLY inefficient!
APK
P.S.=> It's all here in black & white, see subject https://apple.slashdot.org/com... chump... & when YOU DO A BETTER PROGRAM OF THIS NATURE (imitation IS the sincerest form of flattery & you would be imitating me)? Then, talk (you've already TALKED OUT YOUR ASS taking shots @ me & I shot you down for it easily)... apk
The original point is about progressbar updates. You STUPIDLY used Application.Processmessages to refresh it on a SHORT run? ProgressBar1.refresh/.update works better (drops no messages like Application.ProcessMessages can & works DIRECTLY on the control itself using its own methods (less messagepass overhead by FAR)). It's SO short of a process (30 seconds over roughly 6++ million records on sort/dedup) it doesn't NEED LOOPS stupid (slowing it down too) & a simple set of lStringList.Duplicates:= dupIgnore; & then lStringList.Load (filename) does the job FASTER by far! Plus, you STUPID little noob amateur, using a Sleep API call directly (vs. Application.ProcessMessages which abstracts that in the end but does add overhead) would have been better... but "poor lil you", the operation DOES NOT NEED THAT (unless you like going slower) as its only seconds of operation.
BIGGEST MISTAKE YOU MADE dumbfuck? It was about progressbar updates & THAT IS A VCL CONTROL - you later told me "put it on a thread" (it's no background op dumbo)?
THE VCL IS NOT GUARANTEED THREADSAFE (the Win32 api is though) SO YOU TAKE RISKS ON DROPPING DATA (not mere messagepassing updates this time but ACTUAL DATA).
You WEAK little bigmouthed FUCK - you lose & suck!
APK
P.S.=> Above ALL else though you WEAK bitch? Show us you've done a BETTER program of this nature BEFORE I did - fact: YOU CAN'T whimp (you're all FUCKING talk, & hot fucking air) - you fucking punk! apk