New Tesla 2022 revisions S and X receive AMD ZEN1 based processor and NAVI23 GPU
Click here to post a comment for New Tesla 2022 revisions S and X receive AMD ZEN1 based processor and NAVI23 GPU on our message forum
rl66
Not only, due to shortage, as an exemple you can found NVidia or ARM in it...
What is exeptional is that the user experience is the same whatever system you have: it mean also a complete rewriting of the system.
It's also why concurent can't do car while they can because they addapt themself.
fantaskarsef
well, good for AMD
schmidtbag
Well isn't that a giant middle-finger to Nvidia. First Tesla was like "hurr durr we can make our own CPU!" which to my knowledge, didn't really come to fruition. During this process, Nvidia was (surprisingly) cooperative, basically saying they would happily work together again if given the chance. Yet, now they're going with AMD? I don't see a mistake in this decision, but it really gets me to wonder what it is about Nvidia they so desperately want to avoid. The fact it's Zen1 makes it even more confusing, though, I understand the CPU isn't really doing a whole lot for the car; the GPU is where all the "magic" happens.
Embra
Good deal on Zen1's, and stock to boot.
Certainly enough to handle Tesla's needs coming from Arm.
Helps them fill there back log of orders.
tunejunky
the cpu controls hvac, the (elaborate) lighting, the cameras (over a dozen on each Tesla), systems and system updates, and controls the wide variety of sensors.
i can give a bit of local insight (Tesla, AMD, Nvidia within 10 minutes from my front door) and tell you it's a lot to do with "strong" personalities. and not just Elon and Jen-Hsun (his real name), the Nvidia ARM marriage is only now showing the synergy of their efforts - and it's too expensive for a Tesla. Tesla has a tight cost control regime they instituted with the model 3 (which they needed for investors ) as at that time there was cynical doubts about the ability of Tesla to hit target production (which they eventually did).
Zach Kirkhorn is the Tesla "Master of Coin" aka C.F.O .and he's the one you can look to for scotching any deal with Nvidia (who doesn't do cheap or favors).
Venix
To my understanding it has to be that nvidia is a difficult partner to work with , examine ps3 and xbox . The aibs have the fear of god ... And in the past numerous times they threw under the bus even tsmc and other fabs blaming em, i do not know if this is what it is but they seem to be doing business with iron fist so makes sense if companies have an alternative to turn to em. That said obviously what ever nvidia doing is clearly working, mind you i am just speculating here.
rl66
I am not totaly agree with model 3: Tesla needed it for European road, the other model can only do fastlane or autobahn and are totaly inadapted for other road and inner city road.
Even the model 3 is a "big car" here, but it's less expensive than the Renault Zoe (Nissan Leaf) and so successfull due to taxes on fossil combustion engine.
For those interested in why Tesla have many CPU in their computer there was a reportage from 2 or 3 week ago about the car semi conductor crisis and how tesla do to still produce car without cut some part of the car's equipement.
cucaulay malkin
they seem like a company that wants to do business in the "my way or the highway" fashion.
like one burned bridge is an easy sacrifice if you've the resources to build two more in its place.
amd like to make long-term partnerships.nvidia are willing to find the most lucrative short-term ones and then see how they develop.
tunejunky
i hear you, but you are mistaken.
the Model 3 (more of which have been sold in California than all of Europe) was the make or break product for Tesla.
it was where Elon's rubber 😱 met the road. Tesla had made an (near) endless string of pushing back delivery dates and had to engineer an entirely new way of making a car. which ate up the vast majority of their cash on hand and set an automotive record for the use of (short term) contract labor.
i only live in the same town as their (soon to be former) HQ and i'm a (small) investor.
as far as AMD goes the simple fact is that it's easier to get late model apu/cpu's and that they'll do the job just as nicely as bespoke silicon but at a fraction of cost (see above).
icedman
If it's zen 1 they're probably using glofo since they made all of AMDs 12/14nm and it's probably a mcm under the heatspreader with 7nm Navi being used
Imglidinhere
That's a hell of an upgrade from what they were using previously! Man... talk about a ton of additional computational power in your car! o.O
Kaarme
That was my thought as well. According to a documentary I watched a few months ago, GF as well is fully booked, but I'd think they can make room more easily than TSMC or Samsung on their most advanced nodes. Who knows, Tesla might have got such chips from AMD's GF quota, meaning they wouldn't have even needed to bid for anything.
RaulR
If they keep raising those prices, soon a Tesla model S will be the cheapest GPU you could buy...
