Model S/X deliveries with Intel-based MCU

[elderforester]

I CC'd my Owner Advisor in the original email I sent to my Delivery Specialist asking about the new MCU, since my DS did not give me a proper answer. For anyone curious, this is what my OA said:

"Just want to let you know that your Model S should have the new MCU according to the built spec."

Crossing my fingers hoping this is true. Prod started 3/12/18. Status says still under production on My Tesla account.

Specs: 75D, Pearl White exterior, White Interior w/ carbon fiber decor, PUP, 19inch standard wheels, Sunroof, VIN: 2468XX.

Looks like we LA folks lucked out in this event. Yours definitely will have the newest hardware. Mine (Vin 2467xx) had same start date as yours, 3/12. Hope to be collecting it on Thursday 3/22.

Improved features will continue to be designed and installed into the Model S sometime after we take delivery, however. I can live with that.

 

[whitex]

Given the choice, I would have gladly delayed production by a week, instead of possibly refusing delivery and waiting an additional 3 f-ing months.

Anything not made in for March pushes out to June. Would you have gladly pushed out to June? If yes, why not just cancel and re-order now?

What they should have done is used the last 1-2 weeks of old MCU production for inventory cars to sell on the site. Would have drawn out the timeline and pissed off less people.

No matter what they do there will always be people who bought just before or just after. As per your other comments above, if people are happy to delay by a couple of months, maybe the best solution would simply be to allow people to transfer their deposit to a new order (rather than foregoing it). I suspect however that there will always be people who will be unhappy no matter what Tesla does, so realistically there is no good solution.

 

[whitex]

I walk into a store, buy 10 bananas for a dollar.

You walk into the same store 1 minute later, as I’m waiting for my bananas.

Store owner says to you, 20 bananas for a dollar, hands me my 10 bananas. You buy your 20 bananas for a dollar.

Did I know the value of my bananas when I bought them? Yes.

Am I upset because I could have gotten more for my money 1 minute later had I known the adjustment was coming? Absolutely.

A line has since formed for the 20 banana deal. I could have waited a minute and gotten the deal, but now I have to go to the end of the line and wait 3 months for the banana deal.

Do you ever trade stocks? That is exactly what happens. If you trade some volatile and/or low volume stocks, by the time your trade settles the price could be significantly different. The you buy it for $150 but a moment later it's $130!

 

[CarbonFree]

No matter what they do there will always be people who bought just before or just after. As per your other comments above, if people are happy to delay by a couple of months, maybe the best solution would simply be to allow people to transfer their deposit to a new order (rather than foregoing it). I suspect however that there will always be people who will be unhappy no matter what Tesla does, so realistically there is no good solution.

People may be unhappy, but that will always be true of any action--even one which is an improvement for most, if not all. You shouldn't use that as an excuse not to make anything better. It's just a cop out.

Given that the current MCU has been in use for five years, a paid upgrade would likely satisfy 99% of users. Most of those who want to upgrade will have purchased within the last year. 1% may still be unhappy, but 99% of users who want a new MCU will be satisfied. That's much better than leaving 100% of users who want a new MCU dissatisfied.

 

[whitex]

People may be unhappy, but that will always be true of any action--even one which is an improvement for most, if not all. You shouldn't use that as an excuse not to make anything better. It's just a cop out.

Given that the current MCU has been in use for five years, a paid upgrade would likely satisfy 99% of users. Most of those who want to upgrade will have purchased within the last year. 1% may still be unhappy, but 99% of users who want a new MCU will be satisfied. That's much better than leaving 100% of users who want a new MCU dissatisfied.

To satisfy 99% of the user the upgrade would have to be dirt cheap. If Tesla came out today with "you need a new CIM, new IC, new mirrors, new wiring harnesses, new security module, tons of labor to rip out old wiring and add new", total retrofit cost $25K, I guarantee you would not have 99% people happy. This very same discussion took place in the past when the dual motor and AP1 was introduced, then when AP2 came out. LTE upgrade was the closest to making people happy, but only because it costed $500 - still I don't think the percentage of unhappy people was only 1%.

 

[CarbonFree]

To satisfy 99% of the user the upgrade would have to be dirt cheap. If Tesla came out today with "you need a new CIM, new IC, new mirrors, new wiring harnesses, new security module, tons of labor to rip out old wiring and add new", total retrofit cost $25K, I guarantee you would not have 99% people happy. This very same discussion took place in the past when the dual motor and AP1 was introduced, then when AP2 came out. LTE upgrade was the closest to making people happy, but only because it costed $500 - still I don't think the percentage of unhappy people was only 1%.

Tesla recently introduced a Roadster battery upgrade which cost $29,000. And yet, they still did it, and people still bought it. And they knew, before they offered it, that the uptake would be less than a thousand people--satisfying a fewer number of users than an MCU upgrade would.

And yet, I can't help but feel that a $25K number is a scare tactic. If I can buy a third party CarPlay LCD at Best Buy for $250, Tesla can make an upgrade which costs less than $25K. It wouldn't need to upgrade the screen, the mirrors, or any wiring harnesses; all it would have to do is keep existing functionality with a faster processor. Heck, they could take the original MCU box, swap out the Tegra board for an Intel board which ran Tegra ARM code under software emulation, keep all external connectors the same, and it would still be a huge improvement.

Similarly, 1% of dissatisfied users is an example. If the number of dissatisfied users were 5%, 10%, or even 25%, we're still ahead, but the actual calculation depends on many factors, of course.

 

[kirbytazzy]

So I just got the same call unfortunately... VIN 2468XX March 7th build start. My delivery specialist hasn't been able to answer my question on whether or not my car was equipped with the new MCU but common sense would suggest it is... He said there is a bulletin that just came out last night stating that there is an issue with the IC coming on (it stays blank) when opening the door and that Tesla is suspending all deliveries until a fix is out. He states Tesla's engineers are all over it and they are hoping to roll out an update in the next 24-48 hours hopefully. Does anybody have access to these bulletins or have any additional info? Is there really a reason Tesla can't allow us to take delivery as planned until it's fixed? From a customer service standpoint I'd rather take possession of a car I'm paying for on time and possibly expect to experience a "bug" than to delay the delivery when an OTA update is expected either way.

I got the same phone call from store mgr. I was going to take delivery Mon.

 

[Sandollars]

Wonder if this thread can surpass the Model 3 Alcantaragate?

LOL!

Only if the upgrade turns out to be a catastrophic disaster or full of bugs!

 

[whitex]

And yet, I can't help but feel that a $25K number is a scare tactic. If I can buy a third party CarPlay LCD at Best Buy for $250, Tesla can make an upgrade which costs less than $25K.

I gave that as an example on the expensive side. From what we know today, you'd need:
1. New central display unit (was Tegra, now Intel)
2. New instrument cluster (was different Tegra, now seems to run on Tegra)
3. New mirrors with additional antennas
4. New security module (handles keys)
5. New Wifi module (speculation)
6. New wiring harnesses

What if the new display unit or instrument cluster was optimized and requires different mounting, what if it requires larger space behind (want to have a new MCU sticking out 1/2 inch?). Want to add to is a new chasis replacement with proper dash mounts? If you want to just retrofit, now you have to recertify/crash test etc, to make sure its still save (can't just screw in some plastic brackets or drill new holes and hope for the best). What if there were other optimizations/changes?

All these can add up to quiet a bit of parts and labor, not to mention paying someone to design the retrofit procedure and then train all the techs in all SC's - all costs that must be passed down to the customers paying for the upgrade.

Heck, they could take the original MCU box, swap out the Tegra board for an Intel board which ran Tegra ARM code under software emulation, keep all external connectors the same, and it would still be a huge improvement.

LOL, you're obviously not an engineer or at least never designed anything like this (and then had to support it for 15 years) - now that would be EXPENSIVE, and yet offer you no new functionality (say using your phone as a key). If it's just speed you are after, cheaper and faster than designing and optimizing such an "emulation board" would to simply go back and rewrite the Tegra software. the IVI runs on a Tegra 3, which is the same chip you find in the original Microsoft Surface. I have to say, other than lack of Flash support, the UI was very responsive and so was the browser (I still have one, still very responsive). Of course you may say that Windows is somehow faster than Linux but I'll let you fight that battle with plenty of Linux superfans out there.

 

[CarbonFree]

I gave that as an example on the expensive side. From what we know today, you'd need:
1. New central display unit (was Tegra, now Intel)
2. New instrument cluster (was different Tegra, now seems to run on Tegra)
3. New mirrors with additional antennas
4. New security module (handles keys)
5. New Wifi module (speculation)
6. New wiring harnesses

I believe you need exactly one of these things: the first. People here just want their MCU to be faster. They don't care about the speed of the instrument cluster, even. And while everyone would prefer a nicer screen, the screen and MCU are distinct. You can upgrade the MCU without switching out the screen.

If it's just speed you are after, cheaper and faster than designing and optimizing such an "emulation board" would to simply go back and rewrite the Tegra software.

Were this an option, Tesla would have done this instead, rather than going with Intel, as it doesn't require redesigning the MCU part. The Tegra is either too slow, or optimizing the software was more expensive than the port to Intel + MCU redesign.

 

[whitex]

I believe you need exactly one of these things: the first. People here just want their MCU to be faster. They don't care about the speed of the instrument cluster, even.

If you have a Tesla, go see how crippled the instrument cluster is when the large screen is rebooting. It would be in that state if you just replaces the MCU. The new MCU drives the instrument cluster directly (cannot be just connected to old display), so if you just replaced the MCU you'd have an orphaned IC with a message to wait for the main display to boot up (which will never happen since the old MCU is gone) - I bet most people would balk at it.

Also, if you replace just the MCU and not all the other components, now there is a while new configuration that Tesla has to code up, test and maintain for years - all that costs a lot of money.

Were this an option, Tesla would have done this instead, rather than going with Intel, as it doesn't require redesigning the MCU part.

Not at all - of course it was an option, it was just cheaper to throw more CPU power at it than optimize the code (or switch to a new OS).

 

[MarkS22]

It is possible that the Tegra 3 is being end-of-lifed and that the supply of them will come to an end soon. Or it could be that they did the redesign as a cost cutting measure. (Replacing two Tegra systems with a single Intel Atom system?)

I tend to go with the most obvious: It’s a faster processor with better performance leading to a better user experience.

No matter the reason, the faster MCU is welcome.

 

[McSaggler]

Looks like we LA folks lucked out in this event. Yours definitely will have the newest hardware. Mine (Vin 2467xx) had same start date as yours, 3/12. Hope to be collecting it on Thursday 3/22.

Improved features will continue to be designed and installed into the Model S sometime after we take delivery, however. I can live with that.

Yes I agree that future updates will always be rolling out, however the screen was the biggest drawback on the Model S. The screen itself was almost enough reason for me to go through with my Model 3 reservation instead of purchasing an S. Luckily they fixed the issue right before my delivery date so now the S is a much more attractive option than the 3.

 

[MarkS22]

uncorking for nearly all models going back to P85D in early 2015, ludicrous for P85Ds, various roadster upgrades. LTE upgrades. And a bunch I'm not remembering. And no other car mfr comes close. Basically your statement is about as completely wrong as it could possibly be.

You're taking my quote out of context. Your response conveniently cut off the next sentence: "They tend to only include improvements that will also benefit the latest hardware." I’m talking about continued software support for previous hardware versions. What does that have to do with hardware upgrades that you're listing? This is specifically in reference to the discussion on this thread, where we’re talking about Tesla continuing to improve software on hardware that was upgraded.

For example, HW1/AP1 basically stopped development with HW2. I’m suggesting the Tegra MCU will see a similar slowdown in new features/development now that the Intel powered MCU is on all three vehicles. I believe they've squeezed as much as they wanted out of the Tegra and it makes sense for them to support it one chipset moving forward.

 

[dhanson865]

.

You're taking my quote out of context. Your response conveniently cut off the next sentence: "They tend to only include improvements that will also benefit the latest hardware." I’m talking about continued software support for previous hardware versions. What does that have to do with hardware upgrades that you're listing? This is specifically in reference to the discussion on this thread, where we’re talking about Tesla continuing to improve software on hardware that was upgraded.

For example, HW1/AP1 basically stopped development with HW2. I’m suggesting the Tegra MCU will see a similar slowdown in new features/development now that the Intel powered MCU is on all three vehicles. I believe they've squeezed as much as they wanted out of the Tegra and it makes sense for them to support it one chipset moving forward.

Autopilot 1 depends on Mobileye. That hardware would also be in Autopilot 2 if they would have let Elon's plan progress. There is a blank CPU spot on the motherboard for Autopilot 2.0. If it were populated software development for Autopilot would have been considerably faster and could have been done in stages as they moved to newer hardware.

If you want to talk about supporting only software there are tons of examples of that. Non Autopilot cars got the new software that Autopilot cars got. Roadsters got software updates. Autopilot 1 cars are still getting software updates as we speak.

Saying they never updated software for discontinued hardware is simply, categorically, objectively false. It's a straw man argument. You expect anyone that replies to document each case they generalize about too I'm sure (ok I'm guessing). I'd rather go to bed and sleep than to dig up a bunch of old threads about software updates.

 

[chillaban]

For example, HW1/AP1 basically stopped development with HW2. I’m suggesting the Tegra MCU will see a similar slowdown in new features/development now that the Intel powered MCU is on all three vehicles. I believe they've squeezed as much as they wanted out of the Tegra and it makes sense for them to support it one chipset moving forward.

Many people swear that AP1 has been steadily improving even during the period of time when AP2 was floundering. Tesla also delivered smoother 2-car-ahead radar right before AP2 came out, and continues to refine radar braking. If AP1 is the cautionary tale of abandonment, then I'm not losing any sleep as a MCU1 owner. Unlike the case of AP1 when it is an aging single-purpose hardware that has been pushed above and beyond what any other automaker (short of Cadillac lately) has attempted to do with it, I have more belief that the Tegra MCU can continue to deliver meaningful feature enhancements.

 

[MarkS22]

.


Autopilot 1 depends on Mobileye. That hardware would also be in Autopilot 2 if they would have let Elon's plan progress. There is a blank CPU spot on the motherboard for Autopilot 2.0. If it were populated software development for Autopilot would have been considerably faster and could have been done in stages as they moved to newer hardware.

If you want to talk about supporting only software there are tons of examples of that. Non Autopilot cars got the new software that Autopilot cars got. Roadsters got software updates. Autopilot 1 cars are still getting software updates as we speak.

I'm familiar with with DrivePX1 having the EyeQ3. I believe it was the plan all along to build on top of it, but we know that fell apart. It seems we're right about at the point now where we would've been 2 years ago. Much of that EyeQ3 functionality has been ported, so now it's hopefully just more features.

That said, the fact remains the cars with EyeQ3 have not received the updates expected (e.g. traffic light/stop sign recognition, on-ramp to off-ramp, more complex summon, radar whitelisting for more reliable AEB). And for good reason: They're clearly focusing the majority of resources on the new EyeQ-less DrivePX2 platform. Once HW2 was released, AP1's features and improvements dropped off in frequency dramatically. It hasn't received anything significant in terms of EAP functionality for quite some time.

The point being, it doesn't make a lot of sense for Tesla to continue to throw a lot of resources on an older hardware that's a dead-end. AP1 is a hardware dead end (despite the good intentions of keeping an EyeQ as part of HW2.0) and now Tegra is a dead end. I'm not even saying this negatively. They have to manage resources. I'm merely speculating that the Tegra will enter a maintenance mode of upgrades. Any future upgrades/efficiencies that can work on it will be implemented, but they won't go out of their way to optimize for it.

Simply: I think Tesla will spend a lot less time on the Tegra MCU now.

 

[CarbonFree]

Case in point, Tesla implemented an improved spatial enhancement more for audio on the Model 3 MCU; it has yet to appear on my X, even though it's software is several versions ahead. I suspect it will never appear.

 

[alloverx]

Hope MCU2-Intel does not use too much battery, otherwise folks will complain and want to replace it back with MCU1-Tegra.

 

[bhzmark]

You're taking my quote out of context. Your response conveniently cut off the next sentence: "They tend to only include improvements that will also benefit the latest hardware." I’m talking about continued software support for previous hardware versions. What does that have to do with hardware upgrades that you're listing? This is specifically in reference to the discussion on this thread, where we’re talking about Tesla continuing to improve software on hardware that was upgraded.

For example, HW1/AP1 basically stopped development with HW2. I’m suggesting the Tegra MCU will see a similar slowdown in new features/development now that the Intel powered MCU is on all three vehicles. I believe they've squeezed as much as they wanted out of the Tegra and it makes sense for them to support it one chipset moving forward.

AP1 improvements didn’t stop with Ap2 roll out.
All the software updates for pre-AP and AP1 cars isn’t the latest hardware when they got furthe software update support.

And the hardware improvements upgrades are a very obvious kind of support to the legacy hardware since you can only do so much with software support.

Again you couldn’t be wrong to claim that Tesla doesn’t support legacy hardware when they do more hardware and software support of legacy hardware than any car mfr, and really any other mfr of any kind.