BMW iM3 - The Future Electric M3 Super Sedan

[OVERKILL]

As digital transformation continues to change our world, BMW and the rest will have to change, adapt and find their way. Everyone else is playing checkers while Tesla is playing chess. Anyone else's firmware development vertically integrated?

Tech is a differentiate-er, a disrupt-er and quickly becoming requisite. Go big or go home. Who's the big boy now?

I'm not sure the purpose of needlessly inflammatory posts like this one. They really do nothing but poison the well of discourse. It's basically baiting/taunting for those not invested in the "Elon is literally God" mindset.

If the Cyber Truck is Tesla playing chess, maybe they should go back to something a bit more their speed, like eating paste or:

See, I can be needlessly inflammatory too

As I've said repeatedly, the i4 M50 is the nicest EV I've driven (I have not driven a Taycan). If I were in the market, that's what I would buy, the Model 3 and Y weren't in the same class for quality and comfort.

 

[JeffKeryk]

I'm not sure the purpose of needlessly inflammatory posts like this one. They really do nothing but poison the well of discourse. It's basically baiting/taunting for those not invested in the "Elon is literally God" mindset.

If the Cyber Truck is Tesla playing chess, maybe they should go back to something a bit more their speed, like eating paste or:
View attachment 191067

See, I can be needlessly inflammatory too

As I've said repeatedly, the i4 M50 is the nicest EV I've driven (I have not driven a Taycan). If I were in the market, that's what I would buy, the Model 3 and Y weren't in the same class for quality and comfort.

My post was not intend to be inflammatory. I am stating that Teslas are a tech based vehicle, more than any other vehicle, and code their own firmware. Who else has vertically integrated chip technology? I don't know of any.

I salute BMW with their Neue Klasse platform that is no-compromise EV from the ground up and embraces integrated smart technology. Enter the 'Heart of Joy' control unit; that could be a game changer. That's what my thread is about; please refer to post #1 and the thread title.

Right now no one is close to Tesla in this regard. It is great, IMO, to see companies like BMW enter the game; I expect others to do so, and find this fascinating.

 

[OVERKILL]

My post was not intend to be inflammatory. I am stating that Teslas are a tech based vehicle, more than any other vehicle, and code their own firmware. Who else has vertically integrated chip technology? I don't know of any.

Well, how it was written, it came off as needlessly inflammatory

Modern vehicles are all "tech based". They all have screens, everything is electronically controlled. How do you qualify "more than any other vehicle"? Surely integrating multiple screens (tech) with a HUD (more tech) as well as physical buttons uses more tech than just slapping an iPad on the dashboard? How about using more than just cameras for automated cruise control?

The difference here mainly appears to be that Tesla is generally not involving third parties (for things like Android Auto, Car Play, modules from different vendors) when compared to other marques.

You've concluded that this is advantageous, but I can think of several examples in the IT space, such as Sun Microsystems, Cray and DIGITAL, where, while initially advantageous, ultimately were beaten by and either absorbed by their competition or slipped off into oblivion after being out-priced and out-performed by companies like HP, DELL and IBM who were using commodity hardware from a variety of vendors (Foxconn, Intel, AMD, SuperMicro...etc) to produce what were ultimately more successful products.

I salute BMW with their Neue Klasse platform that is EV from the ground up and integrated smart technology. Enter the 'Heart of Joy' control unit; that could be a game changer. That's what my thread is about; please refer to post #1 and the thread title.

Right now no one is close to Tesla in this regard. It is great, IMO, to see companies like BMW enter the game; I expect others to do so, and find this fascinating.

Close to Tesla in what regard though, completely capturing and in-housing their electronics supply chain? I suspect that won't be the case even with BMW's new system, the parts will be produced by somebody else, using chips from somebody else, because that's never been in their best interest to pursue.

Some of the best reactors on the planet were designed and sold by Westinghouse. The turbines are made by somebody else, like GE, the control systems are made by somebody else, who specializes in those systems, the civil work is handled by somebody else...etc.

You ever look under the hood of a piece of Cisco equipment? Do you know how many vendors they use? IOS-XE is based on Linux. Yet another example of the fact that it's the quality and performance of the final product that matters.

HPL produces some of the best oils on the planet. They manufacture none of the components in-house.

I can keep going, the examples are myriad.

 

[JeffKeryk]

Well, how it was written, it came off as needlessly inflammatory.

Modern vehicles are all "tech based". They all have screens, everything is electronically controlled. How do you qualify "more than any other vehicle"? Surely integrating multiple screens (tech) with a HUD (more tech) as well as physical buttons uses more tech than just slapping an iPad on the dashboard? How about using more than just cameras for automated cruise control?

The difference here mainly appears to be that Tesla is generally not involving third parties (for things like Android Auto, Car Play, modules from different vendors) when compared to other marques.

You've concluded that this is advantageous, but I can think of several examples in the IT space, such as Sun Microsystems, Cray and DIGITAL, where, while initially advantageous, ultimately were beaten by and either absorbed by their competition or slipped off into oblivion after being out-priced and out-performed by companies like HP, DELL and IBM who were using commodity hardware from a variety of vendors (Foxconn, Intel, AMD, SuperMicro...ich aetc) to produce what were ultimately more successful products.

Close to Tesla in what regard though, completely capturing and in-housing their electronics supply chain? I suspect that won't be the case even with BMW's new system, the parts will be produced by somebody else, using chips from somebody else, because that's never been in their best interest to pursue.

Some of the best reactors on the planet were designed and sold by Westinghouse. The turbines are made by somebody else, like GE, the control systems are made by somebody else, who specializes in those systems, the civil work is handled by somebody else...etc.

You ever look under the hood of a piece of Cisco equipment? Do you know how many vendors they use? IOS-XE is based on Linux. Yet another example of the fact that it's the quality and performance of the final product that matters.

HPL produces some of the best oils on the planet. They manufacture none of the components in-house.

I can keep going, the examples are myriad.

My post started with "As digital transformation continues to change our world,"

In the case of smart technology, yes vertical integration has proven to be advantageous. Tesla does not make chips; they code their own firmware which enabled them to repurpose existing available chips during the pandemic shutdown. It also allows for no-compromise control of functionality and integration with other components vs piecemeal products that do not talk to each other. Pretty limiting and an upkeep nightmare. By the way, this is what BMW is working on, taking this concept to the next level. I find this fascinating.

I am not talking about Android Auto, Car Play, etc. That's not Digital Transformation. I am talking about the operation of the vehicle and its capabilities. If you refer to post #1, I bolded this:
"This is a controller that has taken the last 20 or 30 years of our experience into a control unit. Everything that is driving-performance related, chassis-control related, propulsion- [and] powertrain-related is now in one integrated control unit,"

You mention tech companies that in many cases were absorbed; this is the natural company and product lifecycle.
My point is Tesla, and now others like BMW, are developing their own firmware to build better vehicles. AFAIK no car company makes chips. This is Digital Transformation in action. It's not about chips per se, rather it is about new advancements and cutting edge use. It's a big deal.

 

[OVERKILL]

My post started with "As digital transformation continues to change our world,"

In the case of smart technology, yes vertical integration has proven to be advantageous. Tesla does not make chips; they code their own firmware which enabled them to repurpose existing available chips during the pandemic shutdown.

OK, but take Apple vs Android. Google doesn't produce hardware, Apple does. Both produce the software for smartphones. Two very successful systems with little in common. Blackberry was totally vertically integrated, where are they now?

It also allows for no-compromise control of functionality and integration with other components vs piecemeal products that do not talk to each other. Pretty limiting and an upkeep nightmare. By the way, this is what BMW is working on, taking this concept to the next level. I find this fascinating.

So you consider almost no physical buttons, no HUD, and everything on a single iPad-esque screen on the plastic dashboard a "no compromise" approach? Because if so, we are in vehement disagreement.

All modules in modern vehicles talk to each other, that's why they are on a BUS network. It's only limiting in terms of the ability to update them OTA, which may or may not be necessary, depending on the function.

I am not talking about Android Auto, Car Play, etc. That's not Digital Transformation. I am talking about the operation of the vehicle and its capabilities.

Those two things are part and parcel with each other. Using CarPlay for navigation for example, how it integrates into the HUD, that's an integral part of the operation of the vehicle and its capabilities. This all dovetails together.

If you refer to post #1, I bolded this:
"This is a controller that has taken the last 20 or 30 years of our experience into a control unit. Everything that is driving-performance related, chassis-control related, propulsion- [and] powertrain-related is now in one integrated control unit,"

Yes, these were, previously, two or three units that talked to each other over the BUS network. Integrating them together, with more powerful hardware for an application where you are doing BMS and other functions not present on an ICE vehicle likely makes more sense, since those systems (particularly the BMS) may need to be updated in the future. That doesn't mean those functions can't be executed with separate pieces of hardware, like they are currently on say the i4 or iX.

You mention tech companies that in many cases were absorbed; this is the natural company and product lifecycle.
My point is Tesla, and now others like BMW, are developing their own firmware to build better vehicles. AFAIK no car company makes chips. This is Digital Transformation in action. It's not about chips per se, rather it is about new advancements and cutting edge use. It's a big deal.

BMW has an extensive history of working with Apple for example, which was part of the development of CarPay, which is why BMW has such excellent CarPlay integration. Per the article, BMW has been doing their own drivetrain software since at least the i8, so this isn't revolutionary as much as it is evolutionary.

My point about mentioning other tech companies, which I thought I made clear, was that just because you have a greater degree of vertical integration doesn't mean you are going to do better in a given space than somebody who isn't. HP had their own version of Unix, DELL never did. How did that play into the relative successes of the companies when compared to say SGI, that had some pretty flash hardware and their own proprietary software as well and are now non-existent?

BMW will do well because their cars are fantastic. While further developing their own software may be beneficial to the final performance of those products, the reality is that they already produce fantastic cars, which gives them a leg-up on Tesla from a purely technical perspective.

Ergo, I find the reference to "Tesla is playing chess while everybody else is playing checkers" ludicrous and needlessly inflammatory. Their approach to materials and interior design feels like the department was headed up by somebody who formerly worked on the J-Body and maybe interned at Fisher-Price. The panel fitment on the Cyber Truck looks like it was handled by Ray Charles. These are not things we have the luxury to just ignore because they have lots of vertical integration.

 

[JHZR2]

What a schnozz.

 

[JeffKeryk]

OK, but take Apple vs Android. Google doesn't produce hardware, Apple does. Both produce the software for smartphones. Two very successful systems with little in common. Blackberry was totally vertically integrated, where are they now?

So you consider almost no physical buttons, no HUD, and everything on a single iPad-esque screen on the plastic dashboard a "no compromise" approach? Because if so, we are in vehement disagreement.

All modules in modern vehicles talk to each other, that's why they are on a BUS network. It's only limiting in terms of the ability to update them OTA, which may or may not be necessary, depending on the function.

Those two things are part and parcel with each other. Using CarPlay for navigation for example, how it integrates into the HUD, that's an integral part of the operation of the vehicle and its capabilities. This all dovetails together.

Yes, these were, previously, two or three units that talked to each other over the BUS network. Integrating them together, with more powerful hardware for an application where you are doing BMS and other functions not present on an ICE vehicle likely makes more sense, since those systems (particularly the BMS) may need to be updated in the future. That doesn't mean those functions can't be executed with separate pieces of hardware, like they are currently on say the i4 or iX.

BMW has an extensive history of working with Apple for example, which was part of the development of CarPay, which is why BMW has such excellent CarPlay integration. Per the article, BMW has been doing their own drivetrain software since at least the i8, so this isn't revolutionary as much as it is evolutionary.

My point about mentioning other tech companies, which I thought I made clear, was that just because you have a greater degree of vertical integration doesn't mean you are going to do better in a given space than somebody who isn't. HP had their own version of Unix, DELL never did. How did that play into the relative successes of the companies when compared to say SGI, that had some pretty flash hardware and their own proprietary software as well and are now non-existent?

BMW will do well because their cars are fantastic. While further developing their own software may be beneficial to the final performance of those products, the reality is that they already produce fantastic cars, which gives them a leg-up on Tesla from a purely technical perspective.

Ergo, I find the reference to "Tesla is playing chess while everybody else is playing checkers" ludicrous and needlessly inflammatory. Their approach to materials and interior design feels like the department was headed up by somebody who formerly worked on the J-Body and maybe interned at Fisher-Price. The panel fitment on the Cyber Truck looks like it was handled by Ray Charles. These are not things we have the luxury to just ignore because they have lots of vertical integration.

Google and Apple are both innovative high tech companies. Both are into far more than smartphones. As you say, Google is not a smartphone company the way Apple is. By the way, doesn't Foxconn make the majority of iPhones? Apple has created an ecosystem vs Google's more open source methodology. Apple's biggest revenue source is the iPhone (but not their highest margin) while Google's is their Search engine and resulting ad revenue, I believe. Both will continue to grow and innovate; both are leaders based in technology. Both are disrupters.

I made no mention of buttons, in fact I am talking about a much different topic. I am talking about the integration of automobile components via a software first approach. Teslas are built around a single central software architecture. My thread is about BMW taking this to the next level. Buttons and iPads are the front end interface to software, they are not the software. However, a digital interface can be programmed to perform more tasks than manual controls, not to mention voice controls which is what I make use of.

"No compromise" refers to a single purpose approach to a product. If a company codes firmware for its own products, there is no need to entertain other requirements. This is a prime benefit of vertical integration. Complete control from design forward. My career was a prime example; I was up against ASK Manman and SAP but had a clear advantage in that I had only 1 customer to satisfy.

No one says BMWs are not great cars, certainly not me. No one would discount their legendary history. BMW's fit and finish is world class. Upscale German painting is as good as it gets. I've not heard anyone say this about Tesla. This is not the topic of my thread, which I stated in the title and 1st post.
My post is speaking to technology in a forward sense. In the short term, distributed outsourcing strategy reduces production costs but in the long term, legacy automakers lose the ability to adapt, innovate and advance technology. From a top level system point, differing product integration can be a nightmare. Smart components do what they do best but may not talk to each other requiring overhead software, not to mention duplicated low level code.
Tesla has literally billions of miles of real world driving data that is being used in AI research and development. Are other companies deeply vested in AI? Perhaps you are more familiar with other company approaches to concepts like the Dojo Supercomputer.

From the vehicle software integration (and consolidation) standpoint, BMW just might leap frog Tesla with their Heart of Joy innovation. Pretty darn cool, advanced and forward looking if ya ask me. While they haven't done so yet I salute their efforts and product direction.
Interesting times ahead, for sure.

I hope this clarifies my position and as always I respect other's.

 

[OVERKILL]

Google and Apple are both innovative high tech companies. Both are into far more than smartphones. As you say, Google is not a smartphone company the way Apple is. By the way, doesn't Foxconn make the majority of iPhones? Apple has created an ecosystem vs Google's more open source methodology. Apple's biggest revenue source is the iPhone (but not their highest margin) while Google's is their Search engine and resulting ad revenue, I believe. Both will continue to grow and innovate; both are leaders based in technology. Both are disrupters.

My point was more the comparison to Blackberry, that, despite having their own firmware/software, weren't able to properly predict, and satisfy, the market. Samsung and LG are both prominent producers of phones, but don't produce their own software. Both source their software from Google and then "personalize" it.

I made no mention of buttons, in fact I am talking about a much different topic. I am talking about the integration of automobile components via a software first approach. Teslas are built around a single central software architecture. My thread is about BMW taking this to the next level. Buttons and iPads are the front end interface to software, they are not the software. However, a digital interface can be programmed to perform more tasks than manual controls, not to mention voice controls which is what I make use of.

"No compromise" refers to a single purpose approach to a product. If a company codes firmware for its own products, there is no need to entertain other requirements. This is a prime benefit of vertical integration. Complete control from design forward. My career was a prime example; I was up against ASK Manman and SAP but had a clear advantage in that I had only 1 customer to satisfy.

I understand that, but I'm speaking in terms of what the end user and customer, perceives as compromises; their experience. If we circle back to Apple, MacOS (originally OS X) and iOS are built on BSD. It's the UI, the ecosystem, the selection of the hardware, that ultimately makes for the experience, which has historically allowed Apple to charge more for what is ultimately commodity hardware (though this has changed a bit recently with the M-series CPU's) in a very flash package that they design and manage the manufacture of. As you note, Foxconn is responsible for iPhone production (being an expert in semiconductor integration and PCB design), but if we look at the legacy of Apple system design, their approach has been an evolutionary path of embodying not only an excellent interface for their software, but integrating high grade materials with a premium aesthetic and excellent QC to really bring the whole package together.

My old Mac Pro here, Foxconn makes the motherboard, AMD makes the video card (and that firmware), Intel makes the NIC, and that firmware, Broadcomm makes the wireless card, and that firmware, Kingston makes the SSD's, and that firmware. Apple makes the OS (based on BSD) that brings that all together for the experience. Not all that different from how modules from different vendors are brought together in a car.

If we contrast this to BMW and Tesla, BMW (and Mercedes) both do an excellent job on the aesthetic, materials and QC front, while Tesla is lacking. Tesla makes very unique, and often polarizing, design decisions, but their QC and materials selections don't bring it together with that "premium" feel you get from the other two. Part of this is how they've chosen to expose the end user to this "software first" approach you reference, leveraging what I assume is a more cost-effective "tablet" interface, rather than providing more conventional tactile controls blended across a variety of interfaces such as clusters, HUD's...etc.

No one says BMWs are not great cars, certainly not me. No one would discount their legendary history. BMW's fit and finish is world class. Upscale German painting is as good as it gets. I've not heard anyone say this about Tesla. This is not the topic of my thread, which I stated in the title and 1st post.

That's my point, they are fantastic cars, the experience aspect of the car, because ultimately, we are talking about cars here, no? If you wanted to keep this specifically about the software, the statement about "checkers and chess" should probably not have made it into the thread.

My post is speaking to technology in a forward sense. In the short term, distributed outsourcing strategy reduces production costs but in the long term, legacy automakers lose the ability to adapt, innovate and advance technology. From a top level system point, differing product integration can be a nightmare. Smart components do what they do best but may not talk to each other requiring overhead software, not to mention duplicated low level code.

I don't agree. This is done every day. Every computer system is a combination of firmware from a variety of vendors brought together under a 3rd party interface. Again, we look at companies like SUN MicroSystems and how they ended up performing in the market relative to say HP or Dell. A higher degree of vertical integration does not guarantee success.

If I buy an HP server, HP has written, or heavily customized, the UEFI or BIOS system, which in turn speaks to firmware from vendors like Intel, Broadcomm, AMD, Qualcomm, Western Digital, Seagate, LSI...etc. We then top this with the user experience which may be provided by Microsoft, RedHat, xBSD, Oracle...etc. And this system will ultimately communicate with other systems over a network using a standard like ethernet.

As I mentioned earlier, cars from legacy vendors leverage a bus network for modules to communicate with each other. All of these modules can be accessed through the OBD port.

So, while I agree that it may be beneficial for BMW (getting back to the topic of this thread) to roll more of these functions into a more powerful single ECM that they write the software for, I think your characterization of those benefits and how this differs from what is in say the i4, is a bit over the top.

Tesla has literally billions of miles of real world driving data that is being used in AI research and development. Are other companies deeply vested in AI? Perhaps you are more familiar with other company approaches to concepts like the Dojo Supercomputer.

And yet, Tesla's intentional avoidance of Lidar and other technologies due to cost, has impeded them in this space. This statement comes off as almost condescending or at least a bit pompous. I think everybody knows your position on Tesla, but when presented like this, without even a nod given to some of the questionable decisions they've made, how does that promote productive engagement?

From the vehicle software integration (and consolidation) standpoint, BMW just might leap frog Tesla with their Heart of Joy innovation. Pretty darn cool, advanced and forward looking if ya ask me. While they haven't done so yet I salute their efforts and product direction.
Interesting times ahead, for sure.

I hope this clarifies my position and as always I respect other's.

And I hope the above clarifies mine. I think this is great for BMW, but as I said, it's the whole package; the user experience, that ultimately matters, and hopefully they can integrate this in a manner that's evolutionary and showcases their legendary "driver-first" approach to chassis and interface, not alienate their customer base. I say this as a former and potential BMW customer who has not found satisfaction in the way Tesla does things.

 

[JeffKeryk]

My point was more the comparison to Blackberry, that, despite having their own firmware/software, weren't able to properly predict, and satisfy, the market. Samsung and LG are both prominent producers of phones, but don't produce their own software. Both source their software from Google and then "personalize" it.

I understand that, but I'm speaking in terms of what the end user and customer, perceives as compromises; their experience. If we circle back to Apple, MacOS (originally OS X) and iOS are built on BSD. It's the UI, the ecosystem, the selection of the hardware, that ultimately makes for the experience, which has historically allowed Apple to charge more for what is ultimately commodity hardware (though this has changed a bit recently with the M-series CPU's) in a very flash package that they design and manage the manufacture of. As you note, Foxconn is responsible for iPhone production (being an expert in semiconductor integration and PCB design), but if we look at the legacy of Apple system design, their approach has been an evolutionary path of embodying not only an excellent interface for their software, but integrating high grade materials with a premium aesthetic and excellent QC to really bring the whole package together.

My old Mac Pro here, Foxconn makes the motherboard, AMD makes the video card (and that firmware), Intel makes the NIC, and that firmware, Broadcomm makes the wireless card, and that firmware, Kingston makes the SSD's, and that firmware. Apple makes the OS (based on BSD) that brings that all together for the experience. Not all that different from how modules from different vendors are brought together in a car.

If we contrast this to BMW and Tesla, BMW (and Mercedes) both do an excellent job on the aesthetic, materials and QC front, while Tesla is lacking. Tesla makes very unique, and often polarizing, design decisions, but their QC and materials selections don't bring it together with that "premium" feel you get from the other two. Part of this is how they've chosen to expose the end user to this "software first" approach you reference, leveraging what I assume is a more cost-effective "tablet" interface, rather than providing more conventional tactile controls blended across a variety of interfaces such as clusters, HUD's...etc.

That's my point, they are fantastic cars, the experience aspect of the car, because ultimately, we are talking about cars here, no? If you wanted to keep this specifically about the software, the statement about "checkers and chess" should probably not have made it into the thread.

I don't agree. This is done every day. Every computer system is a combination of firmware from a variety of vendors brought together under a 3rd party interface. Again, we look at companies like SUN MicroSystems and how they ended up performing in the market relative to say HP or Dell. A higher degree of vertical integration does not guarantee success.

If I buy an HP server, HP has written, or heavily customized, the UEFI or BIOS system, which in turn speaks to firmware from vendors like Intel, Broadcomm, AMD, Qualcomm, Western Digital, Seagate, LSI...etc. We then top this with the user experience which may be provided by Microsoft, RedHat, xBSD, Oracle...etc. And this system will ultimately communicate with other systems over a network using a standard like ethernet.

As I mentioned earlier, cars from legacy vendors leverage a bus network for modules to communicate with each other. All of these modules can be accessed through the OBD port.

So, while I agree that it may be beneficial for BMW (getting back to the topic of this thread) to roll more of these functions into a more powerful single ECM that they write the software for, I think your characterization of those benefits and how this differs from what is in say the i4, is a bit over the top.

And yet, Tesla's intentional avoidance of Lidar and other technologies due to cost, has impeded them in this space. This statement comes off as almost condescending or at least a bit pompous. I think everybody knows your position on Tesla, but when presented like this, without even a nod given to some of the questionable decisions they've made, how does that promote productive engagement?

And I hope the above clarifies mine. I think this is great for BMW, but as I said, it's the whole package; the user experience, that ultimately matters, and hopefully they can integrate this in a manner that's evolutionary and showcases their legendary "driver-first" approach to chassis and interface, not alienate their customer base. I say this as a former and potential BMW customer who has not found satisfaction in the way Tesla does things.

Fair enough, we seem to agree and disagree on points. Certainly the whole package is important and each of us are free to decide what is important. However, the whole package is not the theme of my thread or comments. This seems to be a point of contention or misunderstanding, depending on one's point of view.

My data collection and Dojo comment flows from a statistical analytics standpoint. Without data it's just opinion. You need to gather data, decide what is important and analyze that data; this is a priority at Tesla and certainly not a new one. I believe companies will prosper by determining what data is critical and how to make the most of it. Analytics are a real-world foundation for better business decisions and products, if they are rigorous. It is a competitive advantage.

We disagree on the firmware and integration approach. This is exactly what BMW is doing and is the focus of my thread. I have no information on who will actually do the work, but I find it fascinating. I don't know if BMW has requisite in-house talent, but if not I am willing to bet they will hire some for critical functionality. In my experience, it's the best way if you are serious about results. No one has done what BMW is proposing.

I appreciate your good conversation. Fair minded people can disagree.

 

[OVERKILL]

We disagree on the firmware and integration approach. This is exactly what BMW is doing and is the focus of my thread. I have no information on who will actually do the work, but I find it fascinating. I don't know if BMW has requisite in-house talent, but if not I am willing to bet they will hire some for critical functionality. In my experience, it's the best way if you are serious about results. No one has done what BMW is proposing.

I'm curious as to whether they take the same approach as Apple and take something that's already pretty darn good (QNX, BSD, Linux...etc) and use it as the base for their own heavily customized software, rather than trying to reinvent the wheel. If we knew more about what they already did with the system on the i8, that would give us some insight here.

I appreciate your good conversation. Fair minded people can disagree.

Absolutely!

 

[Ws6]

The jerks ( ceo) have given up on manual transmissions, and went back to pathetic torque converter automatic transmissions versus the much better dct transmission. From the sound of it M3/M4 sales with manual transmission option is still a 25% take rate.

The ZF8 is amazing. I ran an x3M the other day and could hear the thing cracking off upshifts, and it literally didn't even move. Just cracked off upshifts. I drive a vehicle with a 1speed transmission, and relative to me, the X3 didn't even move when it slapped off a shift. Absolutely greased lightning fast, and with none of the BS of a dct.

 

[Ws6]

You're thinking of the E36 M3.

The E30 M3 was promoted for the '88 MY alongside the 325i models that supplanted the eta-engined 325e models starting in '87. That engine briefly carried over in the base model, which lost the 'e'.

The M3 carried a ~$35k sticker, compared to the $28k 325is, which was only down 25 bhp and about half second slower 0-60.

For most folks, the silky six, more comfortable ride and lower sticker price for similar measured performance made a lot more sense.

But the M3, being a homologation model needing to satisfy a 5,000 unit production run, wasn't made for them, it was made to go touring car racing, and was quite successful at it. They eventually produced a bit under 18,000 of them over a five year span, with about 5,000 or so sold in the U.S. The low-mileage '88 that sold for $250k a few years ago is an outlier, but their values are high.

However, back then, BMW NA didn't consider it a success, and were reluctant to import the E36 M3, which would have to be priced too high in its original form. It carried the S50B30 six, with single-VANOS on the intake, producing 286 (European) horses.

BMW enthusiasts, particularly the CCA club members, weren't happy about that decision, and lobbied for the M3 to be brought to the U.S.

They got their wish, but with a compromise -- the U.S. model would get an enlarged version of the M50B25 engines in the x25i models, the 3.0l S50B30US with 240 bhp, not the fancier "Euro" engine, detuned or not. That allowed them to smartly set the sticker on the new '95 model, like the E30 that ended in '91 MY, at the same ~$35k. It turned out to be a hit.

When BMW updated their engines in '96, to improve efficiency and adopt OBD-2, the lineup got displacement bumps, including the M3 with the 3.2l S52B32US. Horsepower figures didn't change, but torque improved.

But again, U.S. buyers could only look enviously at the rest of the world, where the S50B30 evolved into the S50B32, now with double-VANOS on intake and exhaust, and 321 bhp, making the gap even larger than it was before.

It wasn't until the E46 M3, and small run of late M coupes did the U.S. get engine parity again, with the S54B32, giving 333 bhp is American trim.

Probably just as well, because the more complex and highly-tuned "Euro" engines also had higher maintenance requirements, like manual valve adjustments.

Bah, even Hondas have manual valve adjustments. Its not a big deal.

 

[edyvw]

Anyone who says “no physical buttons “ are way forward, is missing what 99% of buyers want and prefer.
Driving vehicle is not to continue video game from home to the road in absence any hobby that requires utilization of muscles.
Driving vehicle is first and foremost about point A to point B. Then we can talk about enjoying driving (where iPads don’t fit whatsoever).
That is why VW abandoned in EU touch screen functions from 2024, that is why Toyota never went that route, that is why Honda after brief experimentation with touchscreen, moved back to physical buttons even before mid cycle refresh on previous Pilot, that is why Hyundai said they won’t even be experiment with it.

As dor integration or performance, BMW is in the game looooooooooooong before Tesla was or Musk decided to see how countries without apartheid look like.

 

[chris719]

Fair enough, we seem to agree and disagree on points. Certainly the whole package is important and each of us are free to decide what is important. However, the whole package is not the theme of my thread or comments. This seems to be a point of contention or misunderstanding, depending on one's point of view.

My data collection and Dojo comment flows from a statistical analytics standpoint. Without data it's just opinion. You need to gather data, decide what is important and analyze that data; this is a priority at Tesla and certainly not a new one. I believe companies will prosper by determining what data is critical and how to make the most of it. Analytics are a real-world foundation for better business decisions and products, if they are rigorous. It is a competitive advantage.

We disagree on the firmware and integration approach. This is exactly what BMW is doing and is the focus of my thread. I have no information on who will actually do the work, but I find it fascinating. I don't know if BMW has requisite in-house talent, but if not I am willing to bet they will hire some for critical functionality. In my experience, it's the best way if you are serious about results. No one has done what BMW is proposing.

I appreciate your good conversation. Fair minded people can disagree.

I don't agree with you on Musk or Tesla necessarily, but I wholeheartedly agree with you on the benefits of vertical integration for software and electronics. I design both and it's plain to see that you're going to get a sub-par product or at least a late/slow development cycle if you constantly send out requirements and RFPs asking Bosch/Siemens/Continental to write all your stuff and then come back and integrate it. Apple's success is largely due to the fact that they control their entire ecosystem. That doesn't mean you can't use off-the-shelf components, but I don't think doing what the automotive guys currently do by buying already-dated libraries and electronics and gluing them together is ever going to get you a cutting edge product from a software and electronics standpoint.

 

[chris719]

The ZF8 is amazing. I ran an x3M the other day and could hear the thing cracking off upshifts, and it literally didn't even move. Just cracked off upshifts. I drive a vehicle with a 1speed transmission, and relative to me, the X3 didn't even move when it slapped off a shift. Absolutely greased lightning fast, and with none of the BS of a dct.

The DCT was still better in terms of absolute performance and feel. More noticeable on downshifts than upshifts. It's a good transmission but it was a cost/simplification move for BMW rather than performance-motivated.

 

[chris719]

I'm curious as to whether they take the same approach as Apple and take something that's already pretty darn good (QNX, BSD, Linux...etc) and use it as the base for their own heavily customized software, rather than trying to reinvent the wheel. If we knew more about what they already did with the system on the i8, that would give us some insight here.

Absolutely!

Having dealt with QNX for many years, I would argue against it being pretty darn good . BMW already uses QNX though. QNX has good salespeople that like to convince you Linux is scary and Germans like pain so it's a winning combination. I've spent the past 10 years designing it out of devices in favor of Linux.

Apple just uses some of BSD userspace, but it's really a non-critical component I'd say. They reached and picked something serviceable, and then built all of their frameworks on top of it.

The main issue is that the major OEMs other than Tesla outsource pretty much everything and have a very old-school "module" mentality. It's a convenient separation of concerns and probably the only way to effectively manage outsourcing everything, but it gives you no architectural flexibility once you've defined it initially.

 

[edyvw]

Having dealt with QNX for many years, I would argue against it being pretty darn good . BMW already uses QNX though. QNX has good salespeople that like to convince you Linux is scary and Germans like pain so it's a winning combination. I've spent the past 10 years designing it out of devices in favor of Linux.

Apple just uses some of BSD userspace, but it's really a non-critical component I'd say. They reached and picked something serviceable, and then built all of their frameworks on top of it.

The main issue is that the major OEMs other than Tesla outsource pretty much everything and have a very old-school "module" mentality. It's a convenient separation of concerns and probably the only way to effectively manage outsourcing everything, but it gives you no architectural flexibility once you've defined it initially.

The question is what is the end product? How does that work for a driver? The argument about AC physical buttons here is voice control.
1. Every time I listen music, I have to stop to do the simple task of moving the temperature 1 or 2 degrees.
2. Try having 2-3 small kids in the car and do that.

People who put emphasis on software over practicality remind me every time of a meeting I have with IT people at my work. Whenever they come excited, I always ask the question: how will you ruin our lives now for the next several years?
Since the introduction of Microsoft Teams to do everything pretty much at work, we found out exactly who has life and who does not. That is what Tesla reminds me of.

So, some demands by a majority of drivers will make some things harder to implement than maybe software engineers think it should be, but then, cars are not created for them, but for the end user.

 

[OVERKILL]

Having dealt with QNX for many years, I would argue against it being pretty darn good . BMW already uses QNX though. QNX has good salespeople that like to convince you Linux is scary and Germans like pain so it's a winning combination. I've spent the past 10 years designing it out of devices in favor of Linux.

QNX has been very popular in embedded applications, I must believe it has been for good reasons, though I could see how it might not be attractive to some people, particularly given the flexibility that Linux offers. I mean, when even Cisco goes that route (IOS-XE)... well.

Apple just uses some of BSD userspace, but it's really a non-critical component I'd say. They reached and picked something serviceable, and then built all of their frameworks on top of it.

Darwin started off very heavily based on FreeBSD and NeXTSTEP IIRC, and the kernel had many FreeBSD components as part of it, but Apple has done some major customization as well. I remember when you could download Darwin by itself, during the early OS X days. It didn't have the Mac GUI, but you could load X and whatever DE you wanted on it. Shipped with an inverted terminal (white background) because: Apple.

The main issue is that the major OEMs other than Tesla outsource pretty much everything and have a very old-school "module" mentality. It's a convenient separation of concerns and probably the only way to effectively manage outsourcing everything, but it gives you no architectural flexibility once you've defined it initially.

Yes, but I guess how much that matters depends on the functions of the modules in question. Ford of course partners with Microsoft for their infotainment and UI stuff, which, love it or hate it, has been reasonably successful. You can't OTA update the whole car, due to what I assume are some limitations in their legacy approach to module selection, capability and communication, but I think if you bake the necessary stuff that would need/benefit from that form of on the fly updating into a more capable main PCM (like the BMS for example) that a hybrid approach could be very successful. We ARE dealing with standards-based communication and ethernet is quite common in cars at this juncture (including Tesla). Of course then we have to start talking about security, and that's not something that's been an overly strong suit for legacy manufacturers.

Like with phones, PC's and tablets, even if you provide multi-generational update support (which most marques do not do at this juncture, I haven't received a UConnect update in years now, but everything continues to work just fine) eventually you run into hardware limitations with the older equipment. I'm curious how that's handled for old Tesla stuff, like first generation Model S and X, do they still get the same software feature tweaks and updates as the current cars, or do they only get "critical updates" like LTS versions of Windows? @JeffKeryk do you know?

 

[chris719]

QNX has been very popular in embedded applications, I must believe it has been for good reasons, though I could see how it might not be attractive to some people, particularly given the flexibility that Linux offers. I mean, when even Cisco goes that route (IOS-XE)... well.

QNX sucks for a few reasons. First, because it's way behind in terms of support. You want a driver or support for some hardware? Probably doesn't exist if it's newer hardware unless you're willing to fork over $100k+ to their BSP team and wait 6 months. Second, it's a true microkernel RTOS, but most people don't even know why they need an RTOS and end up using it in places that it is wholly unnecessary like GUIs and infotainment systems. RTOS just provided bounded latency, not necessarily low or consistent latency. QNX performance actually suffers compared to Linux and Windows in many ways. Third, even though it's an RTOS it doesn't mean you're immune to all sorts of priority inversion or deadlock problems that I've seen in QNX customer applications. For applications where you need true bounded latency like a control loop, you should just run this on a MCU or similar. For applications where you don't, which is most, you're better off with Linux.

There's only a very narrow set of use cases where it's an actual good idea. Most people who buy it do so because they feel comfortable giving money to someone vs using an open-source OS and having to understand and maintain their own OS. QNX marketing is also really good at scaring people in regulated industries into buying it. I would say it's not very popular on the whole honestly, it's only popular in automotive and somewhat popular in medical.

 

[chris719]

The question is what is the end product? How does that work for a driver? The argument about AC physical buttons here is voice control.
1. Every time I listen music, I have to stop to do the simple task of moving the temperature 1 or 2 degrees.
2. Try having 2-3 small kids in the car and do that.

People who put emphasis on software over practicality remind me every time of a meeting I have with IT people at my work. Whenever they come excited, I always ask the question: how will you ruin our lives now for the next several years?
Since the introduction of Microsoft Teams to do everything pretty much at work, we found out exactly who has life and who does not. That is what Tesla reminds me of.

So, some demands by a majority of drivers will make some things harder to implement than maybe software engineers think it should be, but then, cars are not created for them, but for the end user.

We fully agree on the goal being the end product. I just think that the vertical integration helps you get a better product. I manage teams that develop software and electronics and the end product is always much better, or at least, gets to that state cheaper and faster when you control everything in-house. Just look at what Apple's able to do by controlling all software and hardware.

Off-topic but MS Teams is surprisingly bad. Slack + Zoom works so much better. It's a shame that IT people love to pay Microsoft. I thought MS would have learned their lesson after the garbage that was Office Communicator -> Lync -> Skype For Business, but no.