Tesla is developing one of the most powerful artificial intelligence supercomputers worldwide — for good reason

Tesla’s original supercomputer configuration for AI mainly contained 720 nodes, each consisting of eight 80GB Nvidia A100 GPUs, for a total of 5,760 GPUs. A graphics processing unit (GPU) is a specialized electronic circuit designed to manipulate and alter memory to speed up image creation in a frame buffer intended for output to a display device. GPUs are used in cars, but also in embedded systems, mobile phones, personal computers, workstations, and game consoles.

Modern GPUs are efficient in processing computer graphics and image processing. Its parallel structure makes it more efficient than general-purpose CPUs for algorithms that process large blocks of data in parallel. In a personal computer, the graphics processing unit (GPU) can be located on a video card or built into the motherboard.

With the recent addition of another 200 nodes with the same characteristics (1,600 GPUs), Tesla’s AI reaches a total of 920 nodes, which represents 7,360 GPUs. This update makes the block the seventh most powerful supercomputer in the world, Tesla’s director of engineering Tim Zaman explained on Twitter, although this capacity has not yet been publicly evaluated.

The fact that the system cannot be evaluated means that it cannot officially be part of the Top500 of the world’s most powerful supercomputers. If it does, it will run into systems with similar computing power, such as Perlmutter (6,144 Nvidia A100 GPUs) or Selene (4,480 A100 GPUs). However, Tesla is not currently opening the door to allow this, although apart from the above arrangement, the important thing is the benefit that the company can get from this system that precedes a more ambitious one, called “Dojo”, supposedly To be ready by the end of this year.

But, what exactly is the role of the supercomputer inside Tesla? The evolution of autonomous driving systems, at least as suggested by Tesla, requires parallel development in many areas, precisely because what must be achieved is for a car to react like a human – or even better – in any driving situation. Thus, the company is training deep neural networks on problems affecting everything from perception to control through raw images that many of its cars currently circulating on the streets take.

This Tesla supercomputer primarily performs semantic segmentation, object detection, and monocular depth estimation. In digital image processing and computer vision, image segmentation is the process of dividing a digital image into multiple image segments, also known as image regions or image objects (pixel groups). The goal of segmentation is to simplify and/or change the representation of the image into something more significant and easier to parse. Image segmentation is usually used to locate objects and borders (lines, curves, etc.) in images. More precisely, image segmentation is the process of assigning a label to each pixel in the image so that pixels of the same label share certain properties.

In addition, they learn from actual human leadership situations immediately. That way, overall, Tesla can improve the driving assistance systems in its Model Y, Model S and X with updates and pointing in the direction of its long-awaited fully autonomous driving system — the FSD — that Elon Musk has been talking about for a long time. And of course, there’s plenty of other technology built into cars, like machine vision cameras, autonomy algorithms (which models accurate terrain data algorithms in the real world), and FSD chips that power self-driving software.

Tesla Model 3, Courtesy of Tesla Inc.

As we can see, building EVs now and in the future inevitably requires exploring new areas: supercomputing and artificial intelligence are two of the great pillars of Tesla. The company is heavily involved in it, and in addition to promoting the above technologies, it is one of the sponsors of the Conference on Machine Learning and Systems in Santa Clara, California; And on September 30, just like every year, it will celebrate Tesla AI Day, revealing the latest developments in the field and how they are being applied to advance current and future IT plans: We may see Optimus Robot for the first time.

Source: xataka

All images provided by Tesla Inc.

Nico Caballero is the Vice President of Finance for Cogency Power, which specializes in solar energy. He also holds a Diploma in Electric Vehicles from Delft University of Technology in the Netherlands, and enjoys doing research on Tesla and EV batteries. It can be accessed at Tweet embed on Twitter. Nikko covers the latest Tesla and electric car events at Torque News.

Leave a Reply

%d bloggers like this: