Member
- Joined
- Oct 14, 2023
- Messages
- 225
- Thread Author
- #1
China responded to the technological blockade by creating a chip that bypasses all barriers.
Chinese scientists have unveiled a breakthrough in artificial intelligence by creating a chip that outstrips current high-performance AI chips in terms of speed for tasks such as image recognition and autonomous driving. The new chip, called ACCEL (All-Analog Chip Combining Electronics and Light), uses photons for computing and transmitting information, which can significantly speed up the computing process.
In lab tests, ACCEL achieved a computing speed of 4.6 petaflops( PFLOPS), which is 3,000 times faster than the widely used commercial Nvidia A100 AI chip, while consuming 4 million times less power. The discovery could be a significant step forward for China in the race for leadership in artificial intelligence, especially in light of recent restrictions imposed by the United States on access to cutting-edge technologies, including advanced chips.
ACCEL was developed by a Chinese semiconductor Manufacturing Corporation using an outdated 20-year-old transistor manufacturing process, making it more affordable. Researchers from Tsinghua University indicate that the chip's performance can be further optimized by improving the manufacturing process or using more expensive technologies with element sizes less than 100 nanometers.
Unlike semiconductor chips, photonic chips use the physical properties of light, replacing transistors with ultramicroscopes and electrical signals with light. However, the analog computing architecture of the chip limits its application to specific tasks and does not allow you to run various programs or compress files, as do common computing chips in smartphones.
Among the tasks that ACCEL can perform are high-quality image recognition, low-light computing, and traffic identification. The project was funded by the National Key Research and Development Program of the Ministry of Science of China and the National Natural Science Foundation of China.
MakeSens, co-founded by one of the project's researchers, was also involved in the development of the chip. In May, the company released a low-energy chip that uses analog computing.
Dai Qionghai, one of the leaders of the research team from Tsinghua University, said: "Developing a new computing architecture for the era of artificial intelligence is the pinnacle of achievements. However, the more important task is to apply this new architecture to address major national and public needs, which is our responsibility." The research team has not yet responded to questions about the chip's commercial prospects.
Chinese scientists have unveiled a breakthrough in artificial intelligence by creating a chip that outstrips current high-performance AI chips in terms of speed for tasks such as image recognition and autonomous driving. The new chip, called ACCEL (All-Analog Chip Combining Electronics and Light), uses photons for computing and transmitting information, which can significantly speed up the computing process.
In lab tests, ACCEL achieved a computing speed of 4.6 petaflops( PFLOPS), which is 3,000 times faster than the widely used commercial Nvidia A100 AI chip, while consuming 4 million times less power. The discovery could be a significant step forward for China in the race for leadership in artificial intelligence, especially in light of recent restrictions imposed by the United States on access to cutting-edge technologies, including advanced chips.
ACCEL was developed by a Chinese semiconductor Manufacturing Corporation using an outdated 20-year-old transistor manufacturing process, making it more affordable. Researchers from Tsinghua University indicate that the chip's performance can be further optimized by improving the manufacturing process or using more expensive technologies with element sizes less than 100 nanometers.
Unlike semiconductor chips, photonic chips use the physical properties of light, replacing transistors with ultramicroscopes and electrical signals with light. However, the analog computing architecture of the chip limits its application to specific tasks and does not allow you to run various programs or compress files, as do common computing chips in smartphones.
Among the tasks that ACCEL can perform are high-quality image recognition, low-light computing, and traffic identification. The project was funded by the National Key Research and Development Program of the Ministry of Science of China and the National Natural Science Foundation of China.
MakeSens, co-founded by one of the project's researchers, was also involved in the development of the chip. In May, the company released a low-energy chip that uses analog computing.
Dai Qionghai, one of the leaders of the research team from Tsinghua University, said: "Developing a new computing architecture for the era of artificial intelligence is the pinnacle of achievements. However, the more important task is to apply this new architecture to address major national and public needs, which is our responsibility." The research team has not yet responded to questions about the chip's commercial prospects.