In its most recent blog post, NVIDIA, who previously was supplying Tesla, referred to the presentation at the Tesla Autonomy Day investor event.
NVIDIA fully acknowledged the rightness of Tesla's approach to significantly increase the computing power of the on-board computer to be able to achieve full self-driving capability.
"At NVIDIA, we have long believed in the vision Tesla reiterated: self-driving cars require computers with extraordinary capabilities.
NVIDIA noted three main points:
- "Tesla is raising the bar for all other carmakers" - which we assume is great news for NVIDIA as other manufacturers will be looking for a solution to offer competitive system
- "Second, Tesla’s self-driving cars will be powered by a computer based on two of its new AI chips, each equipped with a CPU, GPU, and deep-learning accelerators. The computer delivers 144 trillion operations per second (TOPS), enabling it to collect data from a range of surround cameras, radars and ultrasonics and power deep neural network algorithms." - all the stuff developed in-house
- "Tesla is working on a next-generation chip" - it will be even more powerful, at least three times according to Elon Musk
NVIDIA, however, decided that it must explain and correct "a few inaccuracies in Tesla’s Autonomy Day presentation".
First of all, NVIDIA says it's not right to compare Tesla’s two-chip Full Self-Driving Computer with NVIDIA’s single-chip driver assistance system. NVIDIA DRIVE AGX Pegasus computer for FSD is rated at 320 TOPS (which is more than Tesla's 144 TOPS). The DRIVE AGX Pegasus consist of a two-chip solution, pairing Xavier CPU with a powerful GPU to deliver 160 TOPS - two sets of those delivers a total of 320 TOPS. Moreover, the company is now working on a next-generation processor Orin (in place of the current Xavier).
"It’s not useful to compare the performance of Tesla’s two-chip Full Self Driving computer against NVIDIA’s single-chip driver assistance system. Tesla’s two-chip FSD computer at 144 TOPs would compare against the NVIDIA DRIVE AGX Pegasus computer which runs at 320 TOPS for AI perception, localization and path planning.
Additionally, while Xavier delivers 30 TOPS of processing, Tesla erroneously stated that it delivers 21 TOPS. Moreover, a system with a single Xavier processor is designed for assisted driving AutoPilot features, not full self-driving. Self-driving, as Tesla asserts, requires a good deal more compute."
NVIDIA agrees with Tesla on the massive amounts of computing performance required to achieve full self-driving and says it's ready to deliver similar computers to other manufacturers.
For us, it seems like NVIDIA could become kind of like Google Android (open platform), while Tesla will take the role of Apple iOs (tailor-made solution).
"There are only two places where you can get that AI computing horsepower: NVIDIA and Tesla.
And only one of these is an open platform that’s available for the industry to build on."