Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Update 17.18.50

This site may earn commission on affiliate links.
I took delivery on June 5th and on 17.18.50 as well. No auto high beams and local road limited to 35mph. AP doesn't seem to be as stable as our other model X that's in 17.17.4. Hoping to get silky smooth update this weekend!

Has anyone heard of any updates on expected release of silky smooth? Fingers crossed it comes through this weekend but my past experience with software expected release dates has not been good.
 
I'm not sure this is the right place to ask.

My car, S75D, was delivered on 6/5, we left for a couple of weeks so it sits home in the garage.
I have software 17.18.50. When a new update becomes available, can I use the Tesla app or the My Tesla login on the Tesla site to initiate the installation remotely or does that have to be started locally from the car's own screen?

I see a lot about 8.1 and 17.18.50. Is one software and the other firmware? Is the 17.18.50 a subdesignation of a broader 8.1 category?

Thanks for any help with this.

David
 
I'm not sure this is the right place to ask.

My car, S75D, was delivered on 6/5, we left for a couple of weeks so it sits home in the garage.
I have software 17.18.50. When a new update becomes available, can I use the Tesla app or the My Tesla login on the Tesla site to initiate the installation remotely or does that have to be started locally from the car's own screen?

I see a lot about 8.1 and 17.18.50. Is one software and the other firmware? Is the 17.18.50 a subdesignation of a broader 8.1 category?

Thanks for any help with this.

David

You must start the update from the car, unfortunately. You will get notifications over the app when an update is ready to apply, but for whatever reason, Tesla wants you to be physically inside the car to initiate the update.

As far as 8.0 vs 8.1 vs 17.18.50, the 17.18.50 "build number" is a unique number that implies the version and minor revision. Just like how Windows version "5.1.2600" was the original release of Windows XP, but we just call it Windows XP.

Lately, Tesla seems to be a bit confused about 8.0, 8.0.1, 8.1, and other marketing version designations. For example, Elon referred to an "8.0.1" with enhancements, but we basically got the changes in a build update that still called itself 8.0.


So for people on this forum, we tend to just go by the 17.18.50 number and most of us know what those mean.



P.S. In the current numbering scheme, the first number is the year (e.g. 2017), the second number is the week number (18th week of the year), and the third number is a build number.
 
  • Informative
Reactions: D.E.
I just downloaded & installed 17.24.28. Release note says bug fixes. Going on a trip tomorrow so will see if there's anything new.
I've been driving the new release for about 10 days (including another update to 17.24.30 a couple of days ago) and it might be my imagination, or wishful thinking, but the autopilot (AP1) seems smoother in both lane holding and lane changing.

I was also getting an intermittent "air suspension needs service" warning, primarily at startup, and that seems to have gone away with this release. My SC says it was a firmware bug that was addressed in this release.
 
Glad air suspension bug fixed. But it begs a question: what were they doing futzing about in that code in the first place? Improving something? Fixing some unknown to us bug? We'll never know. But it's interesting.
People reported that the car wakeup became faster, so somebody was working on that, but because there's no integration tests and the "we do not test, but when we do we test it in production" culture, the unfortunate interaction with other car systems that might be taking longer to wake up than the new code in cid was only noticed by users when it was too late (apparently once a new image is deployed there's no rollback at the factory so defective image is pumped up again and again into new cars until there's a fixed one available, I have no idea how it is justified internally).
 
People reported that the car wakeup became faster, so somebody was working on that, but because there's no integration tests and the "we do not test, but when we do we test it in production" culture, the unfortunate interaction with other car systems that might be taking longer to wake up than the new code in cid was only noticed by users when it was too late (apparently once a new image is deployed there's no rollback at the factory so defective image is pumped up again and again into new cars until there's a fixed one available, I have no idea how it is justified internally).
I've worked on test systems for "hardware in the loop (HIL)" testing for Tesla and I can categorically state that Tesla does test the crap out of all the software before it gets installed anywhere. There are very extensive regression tests for all the CAN-bus messages going around in the car. In HIL testing, the engine controller can be tested without the car going anywhere. For example, the test software can send the CAN bus message to the ECU that the accelerator pedal is being pushed, the ECU reacts by sending a message to the drive electronics to increase currents. The test software then measures that current, and having a physical model of the car, computes how much faster the car would be going, it then increases the frequency of a square wave send to the ECU to emulate the wheel tachometer. So the ECU now thinks that the car is going faster. This is a bit simplified but you get the idea.

But still, regression tests don't test for every possible interaction or unexpected or unknown error conditions. You can bet that this particular scenario is now added to the regression tests so it will be caught whenever anybody changes the software again.

Now, testing AP software is a whole other ballgame and I'm not sure at all we'll see any decent Level 4 capability within the next 5 or even 10 years. That's why I did not buy the Full Self Driving Option. As my expectations were set accordingly, I'm still very happy with whatever AP I do have in the car. But I only use it on the highway...
 
I've worked on test systems for "hardware in the loop (HIL)" testing for Tesla and I can categorically state that Tesla does test the crap out of all the software before it gets installed anywhere. There are very extensive regression tests for all the CAN-bus messages going around in the car. In HIL testing, the engine controller can be tested without the car going anywhere. For example, the test software can send the CAN bus message to the ECU that the accelerator pedal is being pushed, the ECU reacts by sending a message to the drive electronics to increase currents. The test software then measures that current, and having a physical model of the car, computes how much faster the car would be going, it then increases the frequency of a square wave send to the ECU to emulate the wheel tachometer. So the ECU now thinks that the car is going faster. This is a bit simplified but you get the idea.
And all of that in just few short hours before deploying to customers! Magic! ;)
(also this is called unit tests)

But still, regression tests don't test for every possible interaction or unexpected or unknown error conditions. You can bet that this particular scenario is now added to the regression tests so it will be caught whenever anybody changes the software again.
Yeah. Hopefully somebody would incorporate a "test it on a real car" mode too (need more than one car obviously, to ensure as many feasible configurations are covered as possible) ;)
Then it would be called integration testing.
 
  • Funny
Reactions: lunitiks