Since last October, Tesla has been equipping all its vehicles with a supercomputer, and Nvidia has also confirmed that it is based on its Drive PX2 platform for autonomous driving. However, the specific product remained unknown, as it comes with different variations.

Nvidia has two versions of the Drive PX2. The 'Lite' Drive PX2 for Autocruise has a single GPU, single camera/radar input, and a single Tegra X2 Parker system on chip (SoC). And the full-strength Drive PX2 for AutoChauffeur features two Parkers, two Pascal discrete GPUs, and several inputs for cameras and radar antennas.

Like Us on Facebook

Most people assumed that Tesla is using the Drive PX2 for AutoChauffeur, but thanks to Model S owner Kyle Day, Tesla's on-board 'supercomputer' has finally been revealed. Of note, Day's teardown of the computer is published in the Tesla Motors Club.

The panel installed on Tesla's vehicles appeared to be a custom board in-between the 'Drive PX2 for Autocruise' and 'Drive PX2 for AutoChauffeur' with one GP106 discreet GPU and one unknown CPU (although it appears to be a Tegra X2 Parker SoC), according to Seeking Alpha.

The recent revelation appears to imply that Tesla sorted for custom solution and its panel is not as powerful as Nvidia's most powerful solution for self-driving. However, Electrek noted that Tesla kept its door open for upgrade opportunities, even announcing its second generation autopilot with the new hardware. Tesla is also allegedly developing its own custom SoC for self-driving cars and will potentially tap Samsung to manufacture.

Meanwhile, Tesla advertised that all its vehicles is powered by the Nvidia Drive PX2 AI computing platform, which could deliver more than 40 times the processing power of its former system and runs a neural net for vision, sonar, and radar processing.

Nvidia, on the other hand, describes its Drive PX2 as the first AI supercomputer for self-driving cars in the world, with a computing power equivalent to about 150 MacBook Pros.