Shulou(Shulou.com)11/24 Report--

Finally, a lot of cursing RTX 40 series graphics card has come, as the latest work of NVIDIA, its performance, not to mention the high price has caused the dissatisfaction of the majority of consumers. However, in the comments section of this article, please do not rush to scold Huang Renxun, because today's protagonist is not the graphics card, but the car chip-Thor (Thor).

NVIDIA entered the field of smart driving very early, and previously launched Orin, which has the computing power as high as 275TFLOPS. According to NVIDIA's original plan, Altan, an iterative product of 1000TFLOPS's computing power, will be mass-produced between 2024 and 2025, but now NVIDIA can't wait and the launch of Thor,Altan has been cut.

Image source: pixabay

In terms of parameters, the Thor is very strong, with a full 77 billion transistors and a computing power as high as 2000TFLOPS. If you don't know what this math power represents, then Xiao Lei, for example, Tesla's Model S, which can run 3A games, has a computing power of 10TFLOPS on a car AMD chip, a gap of 200x. This is not to say that strong arithmetic means better performance, but a gap of 200 times is enough to smooth out many problems in software optimization and adaptation.

Of course, such a powerful computing power is not only to support the operation of the system, but also not only for intelligent driving, but also for intelligent driving. Thor is a multi-functional vehicle chip, which can not only shoulder the performance requirements of intelligent driving, entertainment and other aspects, but also intelligently adjust the distribution of computing power. NVIDIA has a good idea of solving multiple problems with one chip, making it easier for car design and integration, but can this solution really be achieved?

01 all the functions of one chip, is it possible? In order to make the various functions do not interfere with each other, many functions of the car are not the same chip, such as autopilot and entertainment functions, which are two different chips, and the radar detection functions required for autopilot also need to be completed by separate chips. Due to the large size of the car, so the requirements for internal space is not high, car companies will not care about the problem of too many chips, the most is more troublesome design.

NVIDIA's solution is different, to integrate many functions into a single chip, such as Thor, which can be used to run the car system, support functions such as entertainment and navigation, as well as autopilot. This raises the question of whether it is possible for a chip to integrate all the functions needed by a car in the future.

Image source: pixabay

In recent years, many chip manufacturers are engaged in functional integration to increase chip integration. For example, Desai Xiwei is trying to integrate driving assistance and automatic parking on the same chip, and many car companies have chosen this solution. Infineon, a semiconductor company, published an article saying that the increasing integration of automotive computer control modules (ECU) can keep or reduce the size of circuit panels, reduce the number of chipsets and reduce the cost of on-board chips.

However, it is not an easy task to integrate all the functional operations into one chip. How to connect and deal with various sensors in the car is a headache for chip companies. Most likely, these sensors are not produced by the same company.

Therefore, from the technical level, functional integration can be achieved, and the relevant semiconductor companies are working in this direction, but there is still a long way to go to really achieve this step. At present, most semiconductor companies, like NVIDIA, can only integrate one or two functional operations into one chip.

Image source: pixabay

NVIDIA is a special one, because the integrated functions of Thor are the most important automotive systems and autopilot in the future, and the computing power of Thor reaches 2000TFOPS. According to the data of Yiou think tank, for every level of autopilot, the computing power needs to be increased tenfold. In the future, the mainstream L4 autopilot needs 400TFOPS computing power. Thor can allocate some of its performance to maintain other functions of the car under the premise of meeting L4 autopilot.

As of today, Thor is the most powerful chip in the automotive industry, coupled with the influence of NVIDIA, it can be predicted that Thor will become popular in the automotive industry after mass production, and even many car companies will take Thor as the main selling point.

02 Thor, do you want the sparrow to fly up the branch and turn into a Phoenix? From the perspective of math, you may have guessed that the price of Thor will never be cheap. Xiao Lei inquired about the 275TFOPS computing chip Orin that NVIDIA launched before, and the purchasing cost of the car company is about $500 (about 3500 yuan) each, and the price of Thor must be more exaggerated.

The current situation of new energy vehicles is that it is difficult for the old car companies to make a U-turn, and the market capitalization of the new car-building forces with few product sales is approaching the old car companies. Although these new car companies have a unstable foundation, almost all of them have to enter the high-end market in order to promote brand awareness and improve brand quality.

Image source: pixabay

It will not be so easy to hit the high end, but unlike established car companies, many of the new car-building executives come from Internet companies, or with capital injection from Internet companies, and marketing is their best. Established car companies also have publicity, but it is difficult for ordinary people to understand the obscure knowledge of cars, and few people make it clear to ask anyone on the street what is the difference between turbocharging and natural aspiration.

In contrast, the car chip is a very simple publicity gimmick, after all, go to the street to find someone to ask him whether to choose 2000TFOPS or 10TFOPS chip without considering the price. I believe that most people will choose 2000TFOPS chip. In fact, car companies have begun to market the concept of "running points". The direction of their marketing is computing power, that is, 1TFOPS can run 1 trillion times per second.

However, the imagination is very plump, showing a very bony feeling, the blessing that Thor can bring to car companies is very limited, for three reasons: first, the brand image of car companies themselves is insufficient. If the car is not equipped with high-end, it can rush to the high-end market, and the brand value needs to be stable for a long time.

Image source: pixabay

Second, car companies and consumers need to consider costs. Cars use high-end chips, the price will not be low, if the brand value is not recognized by consumers, car sales will certainly not be high, the loss will be more serious.

Third, the ability of integration is insufficient. Chips can solve math problems, but it is impossible for an ordinary car company to instantly have excellent self-driving and auxiliary driving algorithms, let alone provide a driving experience comparable to that of luxury cars immediately. Integration is a very complicated thing, not all car companies can handle it.

There are many new car-building forces that have hit the high-end market in recent years, but only a few car companies, such as Weilai, ideal, Xiaopeng and so on, have achieved good results. The Baiteng car, which was once favored by many investors, lost 8.4 billion yuan. Another car company with strong momentum is Cyrus. With the help of Huawei, it has done well, but it is unclear how far it will go in the future.

For the car companies that have reached the high-end level, Thor is the key to make the product power further, but for the car companies that have failed to reach the high-end, it is the right way to continue to be cost-effective.

03 computing power, should not become a gimmick for car companies, mobile phones, computers and other digital products, in the performance competition often uses a variety of running software, or run some games to compare frame rate and stability. Different cars, self-driving technology has not been fully popularized, and the ecology of automobile software is relatively simple, which is not enough to support the comparison of competitive products of car companies.

Therefore, the concept of computing power, which is often mentioned by virtual currency miners, has become a new gimmick in the mouth of car companies. In some automotive media platforms, Xiao Lei has seen a lot of car math comparison articles, as well as car math rankings and so on. According to the data released by AutoLab, the most powerful car is the Wei Mocha DHT-PHEV lidar version, followed by the Xilai ES7.

Admittedly, the improvement of computing power will greatly improve the performance of some aspects of the car, but for self-driving, user experience, computing is not the most critical, enough, the most important thing should be ecology. When it comes to ecology, we have to blow Tesla.

In addition to the traditional car entertainment function, Tesla unexpectedly wants to move the game console to the car. Model S has already supported the operation of some games, and later Tesla even plans to move the Steam platform to the car. Apple CEO Cook has said that the car is the ultimate form of mobile devices, this ultimate device is definitely not enough to rely on chip computing power, ecology is definitely a more important factor.

Blindly competing with chip computing power will only increase costs and product prices, a situation that consumers absolutely do not want to see. Improving the ecology is the place that car companies need to pay attention to. As a matter of fact, some car companies are already working hard in this respect, for example, Weilai is also "moving" games to cars.

The ecology of cars is still weak, but the new power of car-building has always had a close relationship with Internet companies, and the degree of perfection of car ecology may be faster than we thought. at that time, it will be the time when we really need high-power chips.

This article comes from the official account of Wechat: Feng Xingxing (ID:fengchuxing2021), author: lost Soul.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.