In a recent series of tweets, Elon Musk shared a number of updates on the upcoming rollout of Tesla’s homegrown HW3 chip, which would help the company’s fleet of vehicles achieve Full Self-Driving. According to Musk, the new chip would likely be installed in all new production cars in about 6 months.
Musk noted that Software V9 saw a ~400% increase in useful ops/sec due to the enabling of the system’s integrated GPU and optimizations in the use of the discrete GPU. These numbers, while impressive, are set to be dwarfed by the improvements that Hardware 3 would bring. According to the CEO, the improvements that would be offered by Tesla’s custom chip to the company’s Neural Network would be “somewhere between 500% and 2000%” compared to the company’s existing hardware.
Keeping Tesla’s tendency to continuously innovate, Musk later stated that HW3 would be equipped in all new production cars in around 6 months. Transitioning to the new hardware will not change any of the vehicle’s sensors or production, either, as it is simply a replacement of the Autopilot computer being installed on all vehicles today. As an added note, Musk pointed out that Tesla owners who purchased Full Self-Driving would be given the HW3 upgrade free of charge. Owners who have not ordered Full Self-Driving, on the other hand, would likely pay around $5,000 for the FSD suite and the new hardware.
~6 months before it is in all new production cars. No change to sensors. This is simple replacement of the Autopilot computer. Will be done free of charge for those who ordered full self-driving.
— Elon Musk (@elonmusk) October 16, 2018
Elon Musk noted back in 2016 that all Teslas produced from the Fremont factory are equipped with the necessary hardware – cameras, ultrasonic sensors, and an onboard supercomputer – that would enable Full Self-Driving sometime in the future. Tesla has since rolled out improvements to its hardware since then, though, and in the past Q2 2018 earnings call, Musk and the company’s executives discussed its custom chip further.
During the call, it was revealed that Tesla’s custom chip, which will be specifically designed for self-driving, is being designed by a team led by Pete Bannon, who used to work at Apple, where he helped create the company’s first ARM 32-bit processor that went into the iPhone 5, as well as the first ARM 64-bit processor in the world which went into the iPhone 5S.
“It’s an incredible job by Pete and his team to create this, the world’s most advanced computer designed specifically for autonomous operation. And as a rough sort, whereas the current NVIDIA’s hardware can do 200 frames a second, this is able to do over 2,000 frames a second and with full redundancy and fail-over. It costs the same as our current hardware and we anticipate that this would have to be replaced, this replacement, which is why we made it easy to switch out the computer, and that’s all that needs to be done. All the connectors are compatible, and you get an order of magnitude, more processing and you can run all the cameras at primary full resolution with the complex neural net.”
Tesla appears to be setting the stage for the rollout of its Full Self-Driving suite. At the end of Q3, for one, a leaked email from Elon Musk to Tesla workers revealed that the company is starting an incentivized Full Self-Driving beta test program for employees. In the email, Musk noted that employees who wish to be part of the program would be receiving Autopilot, Full Self-Driving, and the Premium Upgrades Package (PUP) on their vehicles for free, provided that they share 300 to 400 hours of real-world driving feedback. In his message, Elon Musk was optimistic that the ~100 slots for the incentivized beta tester program would be filled quickly.