More than GPT-4! China launched the world's first AI fully automatic design CPU

If you were faced with an answer to a knowledge question, who would you turn to for help? Now, most people would choose to ask GPT. People are gradually getting used to the convenience that technology brings to life.

In this era of AI technology, more and more intelligent AI devices appear in our lives. Researchers at New York University's Tandon School of Engineering created a chip using GPT-4. Some time ago, the Institute of Computing Technology of the Chinese Academy of Sciences and other institutions used AI technology to design the world's first fully automatic CPU chip without human intervention. This also shows that our AI has entered new areas.

1. The birth of Enlightenment One

?Enlightenment One is based on the BSD (Binary Speculation Diagment) binary guess graph algorithm. With AI, the researchers upgraded the original "input and output" to a CPU that could be automatically generated so that engineers did not need to provide any code or natural language descriptions. Removes the logical design and verification links in the traditional design process.

The CPU uses a 65nm process with a frequency of 300MHz and can run the Linux operating system. Its performance is comparable to the Intel 80486SX, and the design cycle is reduced to 1/1000.

2. Enlightenment One VS GPT-4

The most used generative AI in the world is GPT-4, so what is the difference between the chips designed by Enlightenment One and GPT-4?

Previously, researchers at New York University's Tandon School of Engineering used GPT-4 to design a microprocessor chip with simple English conversations. However, this chip is relatively simple, using a 130mm process, and needs more practical application.

Compared with the times, the Enlightenment 1 processor designed by the Chinese Academy of Sciences is designed based on the BSD binary guessing graph algorithm, adopts a 65nm manufacturing process, and generates 4 million logic gates in only 5 hours, which is 4,000 times larger than the chip that GPT-4 can design.

3. Prediction of the future of AI chips

With the continuous advancement of AI technology and the rapid development of AI chips, countries have different directions for the future development of AI.

• AI chip + intelligent driving: With the continuous development of autonomous driving technology, the demand for AI chips will also increase significantly. Different road conditions emergencies need powerful technical algorithms and prediction algorithms to support, and AI chips can come to bring a "super brain" to the vehicle.
• AI chip + medical field: This field may also become a key area for the development of AI chips in the future. AI technology can be applied to disease diagnosis, drug research and development, etc., which can improve medical efficiency and treatment effects.

• AI chip + design: AI chip also has great development potential in this field. With the further maturity of AI technology, the things it has created are relatively perfect, and it has a strong "knowledge reserve" compared to people; it is indeed "smarter."
• AI chip + edge computing: Edge computing refers to pushing data processing and computing closer to data sources and terminal devices. With the rapid development of the Internet of Things, more and more devices need to process data locally and make intelligent decisions rather than relying on cloud computing. In order to meet the needs of edge computing, AI chips for low power consumption and high performance are required.
• AI chip + neural network: As deep learning models continue to evolve, the number and complexity of neural networks are also increasing. More complex models require more computing power for training and inference. Therefore, AI chips need to have higher computing performance and more storage capacity to support these complex neural networks.
• AI chip + data security and privacy: In AI applications, a large amount of data needs to be transferred and processed between devices and the cloud. In order to protect the data security and privacy of users, it is necessary to add security-related hardware modules and algorithms, such as hardware encryption and secure computing, to the AI chip.

• AI chip + energy saving: The energy efficiency of AI chips is an important consideration. Current AI chips are less energy efficient and generate more heat when computing at high loads. In order to solve this problem, future AI chips need to achieve higher energy efficiency and energy-saving effects through new architecture design and process optimization.

4. Will AI chips replace engineers?

Under the influence of science fiction, people have different ideas about AI technology. Some people like it very much, and some people worry about it.

So, will AI chips really completely replace engineers? AMD CTO Papermaster's opinion is no. Although AI has a "super brain," there are still flaws in practice that need to be solved by engineers. Artificial intelligence is always built by programs; it cannot replace engineers, but it can be a powerful adjunct to them.

With the continuous advancement of technology, the application field of AI chips will be further expanded. Do you have any expectations for AI chips? Feel free to leave your thoughts in the comments section.