Tesla has unveiled its latest version of its Dojo supercomputer and it’s apparently so powerful that it tripped the power grid in Palo Alto.
Dojo is Tesla’s own custom supercomputer platform built from the ground up for AI machine learning and more specifically for video training using the video data coming from its fleet of vehicles.
The automaker already has a large NVIDIA GPU-based supercomputer that is one of the most powerful in the world, but the new Dojo custom-built computer is using chips and an entire infrastructure designed by Tesla.
The custom-built supercomputer is expected to elevate Tesla’s capacity to train neural nets using video data, which is critical to its computer vision technology powering its self-driving effort.
Last year, at Tesla’s AI Day, the company unveiled its Dojo supercomputer, but the company was still ramping up its effort at the time. It only had its first chip and training tiles, and it was still working on building a full Dojo cabinet and cluster or “Exapod.”
Now Tesla has unveiled the progress made with the Dojo program over the last year during its AI Day 2022 last night.
The company confirmed that it managed to go from a chip and tile to now a system tray and a full cabinet.
Tesla claims it can replace 6 GPU boxes with a single Dojo tile, which the company claims costs less than one GPU box. There are 6 of those tiles per tray.
Tesla says that a single tray is the equivalent of “3 to 4 fully-loaded supercomputer racks”.
The company is integrating its host interface directly on the system tray to create a big full host assembly:
Tesla can fit two of these system trays with host assembly into a single Dojo cabinet.
Here’s what the Dojo cabinet looks like closed and opened:
That’s pretty much where Tesla is right now as the automaker is still developing and testing the infrastructure needed to put a few cabinets together to create the first “Dojo Exapod”.
Bill Chang, Tesla’s Principal System Engineer for Dojo, said during
“We knew that we had to reexamine every aspect of the data center infrastructure in order to support our unprecedented cooling and power density.”
They had to develop their own high-powered cooling and power system to power the Dojo cabinets.
Chang said that Tesla tripped their local electric grid’s substation when testing the infrastructure earlier this year:
“Earlier this year, we started load testing our power and cooling infrastructure and we were able to push it over 2 MW before we tripped our substation and got a call from the city.”
Here’s how the Tesla Dojo Exapod looks like opened and closed:
Tesla released the main specs of a Dojo Exapod: 1.1 EFLOP, 1.3 TB SRAM, and 13 TB high-bandwidth DRAM.
The company used the event to try to recruit more talent, but it also shared that it is on schedule to have its full first cluster or Exapod, in Q1 2023.
It is currently planning to have 7 Dojo Exapods in Palo Alto.
Why does Tesla need to Dojo supercomputer?
It’s a fair question. Why is an automaker developing the world’s most powerful supercomputer? Well, Tesla would tell you that it’s not just an automaker, but a technology company developing products to accelerate the transition to a sustainable economy.
Musk said it makes sense to offer a Dojo as a service, perhaps to take on his buddy Jeff Bezos’s Amazon AWS and calling it a “service that you can use that’s available online where you can train your models way faster and for less money.”
But more specifically, Tesla needs Dojo to auto-label train videos from its fleet and train its neural nets to build its self-driving system.
Tesla realized that its approach to developing a self-driving system using neural nets training on millions of videos coming from its customer fleet requires a lot of computing power. and it decided to develop its own supercomputer to deliver that power.
That’s the short-term goal, but Tesla will have plenty of use for the supercomputer going forward as it has big ambitions to develop other artificial intelligence programs.
FTC: We use income earning auto affiliate links.More.
The State of Michigan has announced a new partnership with wireless EV charging specialist Electreon and Commercial EV manufacturer Xos, Inc. Together, the companies have secured funding from the state to expand wireless charging availability for commercial vehicles, including UPS trucks in Detroit.
It’s been nearly a year since the Michigan Department of Transportation (MDOT), the City of Detroit, and Electreon ($ELWS) unveiled the first wireless EV charging roadway in the United States.
Michigan installed Electreon’s wireless inductive-charging coils (seen below) on 14th Street in Detroit between Marantette and Dalzelle streets to charge EVs equipped with Electreon receivers as they drive on the road.
At the time. MDOT and Detroit officials said the road would be used to test and perfect Electreon’s wireless EV charging technology in a real-world environment before “making it available to the public in the next few years.”
While the public will not be able to take advantage of wireless charging just yet, commercial EVs are gaining access thanks to a new partnership between Electreon, Xos ($XOS), and the State of Michigan.
Michigan progresses as US wireless EV charging leader
When Michigan announced the first wireless EV charging road in the US last year, officials shared hopes that the Great Lakes State and the city of Detroit could become leaders in the innovation and and deployment of such nascent technology.
Following a press release from Electreon, the State of Michigan confirmed details of the new partnership, which now includes commercial EV developer Xos, Inc. as well. The new commercial partnership is supported by $200,000 in funding from the Michigan Mobility Funding Platform (MMFP), building off the state’s “Make it in Michigan: economic strategy, developed by the Michigan Economic Development Corporation (MEDC) to invest in the state’s people, places and projects.
Through the partnership and coinciding state funding, Electreon will extend its wireless EV charging network and use cases in Michigan. Additionally, Electreon’s technology will be integrated into delivery step vans from Xos in order to “demonstrate wireless charging technological value and its potential to lower the total cost of ownership (TCO) in the electrification of commercial truck fleets.” Stefan Tongur, vice president of business development for Electreon, elaborated:
We’re excited to demonstrate how Electreon’s technology can optimize electric fleet usage and showcase the seamless integration of wireless charging into daily fleet operations, minimizing downtime and enabling charging across time and location. We’re proud to do this work in Michigan, a state fostering innovation and sustainable transportation solutions.
In addition to expanding wireless charging on Detroit’s first “electric roadway,” the Michigan project will enable the installation of stationary wireless charging at a UPS facility in Detroit. Xos co-founder and CEO Dakota Semler also spoke:
We are proud to partner with Electreon and support UPS to demonstrate the potential of wireless charging in commercial fleets. This innovative approach will revolutionize how we power our electric vehicles and drive fleet electrification forward.
FTC: We use income earning auto affiliate links.More.
The iconic British luxury automaker is undergoing a major brand overhaul. With its official debut around the corner, Jaguar’s electric 4-door GT, the first of its new series, was spotted testing on British roads. The new images give us a closer look at what we can expect from the revamped Jag brand.
Jaguar’s new electric GT makes its first appearance
After building internal combustion-powered sports cars for over 75 years, Jaguar will become an all-electric luxury brand from 2025.
The company announced earlier this year that it will start fresh with an entirely new range of EVs. After killing off the F-Type, E-Pace, XF, and soon the F-Pace SUV, we are finally getting our first look at what the new branding will look like.
Jaguar’s new electric 4-door GT was caught testing on British roads. The camouflaged prototypes reveal a radically different look than the Jag models we are accustomed to.
You can see one of the biggest changes is the low-riding, extended silhouette, as opposed to the crossover SUV and sedan models like the F-Type and I-Pace, Jaguar’s first EV.
The front and rear bumper designs also appear much more aggressive and bold than previous models.
Jaguar’s electric GT is being put through the paces ahead of its debut. It has already completed tens of thousands of testing miles (virtual and real-world) and will soon hit public streets worldwide.
The new model will be built in Solihull, UK, where Jaguar recently ended production of its gas-powered models.
It will be the first to ride on Jaguar’s new JAE (Jaguar Electric Architecture), which will underpin its upcoming lineup of high-end luxury EVs with prices over £100,000 ($130,000). The electric GT will have a range of over 434 miles (700 km) and upwards of 575 hp, making it Jaguar’s most powerful car of all time.
Jaguar will debut its Design Vision Concept at Miami Art Week on December 2, 2024. Next Summer, it’s expected to make its official global debut ahead of deliveries in 2026.
What do you think of Jaguar’s new design based on what’s shown? Are you excited about the brand overhaul? Let us know in the comments below.
FTC: We use income earning auto affiliate links.More.
Tesla has pushed a new (Supervised) Full Self-Driving update with the promised end-to-end neural networks for highway driving.
However, it’s only for newer vehicles.
“End-to-end” is what Tesla refers to as neural net-powered AI driving the vehicle from vision to controls rather than the controls being explicitly coded. It’s already the case in all widely released versions of (Supervised) Full Self-Driving (FSD) for city driving, but not for highway driving, which uses another software stack.
Tesla originally planned to deliver it for highway driving in October, but it was only delivered to a small number of vehicles.
In its latest AI roadmap, Tesla said that it would come the first week of November instead.
Now, Ashok Elluswamy, Tesla’s head of self-driving and AI, said that the latest release with end-to-end highway driving (v12.5.6.3) has been widely pushed to HW4 vehicle owners:
With the latest release (v12.5.6.3), FSD is using end-to-end neural networks for driving across highways, city streets and parking lots, and has now shipped widely for AI4 vehicles. Highway driving should be smoother, more natural and even safer than the previous explicit control stack. Check out the different driving styles to set speed and lane change preferences. Enjoy and let Tesla AI know if you have any feedback.
However, there’s no word for the millions of HW3 vehicle owners.
In fact, the only thing promised to HW3 vehicles, which Tesla now called AI3, in its last roadmap is this:
Improved v12.5.x models for AI3 city driving
As we have often reported this year, Tesla has reached the limits of the HW3 computer and now needs to optimize the code with every release despite still being far from its promise of unsupervised self-driving.
Electrek’s Take
This is annoying cause I could really use end-to-end on my HW3 car. I am on v12.5.4.2 and it has been a regression from v12.5.4.1 for me, especially on highways.
Yesterday, it almost drove me off-road when taking my highway exit, which is always a bit shaky because it is a short exit and FSD often swings itself into it. It’s a bit awkward, and my girlfriend never likes it, so I disengage FSD before taking the exit when she was with me, but this time, she wasn’t, and I had the new update.
It again swung left before going right into the exit, but this time, it went way too far, and I was in the shoulder by the time I took control.
I took this exit hundreds of times with FSD and it’s the first time it did that.
I am starting to think we won’t see much more improvements to FSD with HW3 cars and there’s no retrofit computer in sight.
FTC: We use income earning auto affiliate links.More.