[ad_1]
It is artificial intelligence now it helps design computer chips, including the most powerful ones needed to run AI code.
Sketching a computer chip is complex and complicated, and designers have to arrange billions of components on a surface that is smaller than a fingernail. The decisions at each step can affect the final performance and reliability of the chip, so the best chip designers rely on years of experience and knowledge to design circuits that achieve the best performance and power efficiency from nanoscopic devices. Efforts to automate chip design for several decades have been considerable.
But recent advances in AI have made it possible for algorithms to learn some of the dark arts involved in chip design. This should help companies make stronger and more efficient plans in much less time. It is important that the approach can help engineers design AI software by experimenting with different code adaptations along with different circuit designs to find the optimal configuration for both.
At the same time, the rise of AI has sparked new interest in all kinds of new chip designs. Cutting-edge chips are becoming increasingly important for almost every area of the economy, from cars to medical devices to scientific research.
Chips, among others Nvidia, Google, and IBM, are all AI testing tools that help organize components and cables into complex chips. The approach may shake up the chip industry, but it can also lead to new engineering complexities, as the types of algorithms that are being deployed can sometimes have unpredictable ways.
Nvidian, chief research scientist Haoxing “Mark” Ren how to test an AI concept called reinforcement learning it can help organize the components into a chip and how to connect them. The approach that allows a machine to learn from experience and experimentation has been central to some of the major advances in AI.
Ren’s AI tools in simulation simulate different chip designs, training a large artificial neural network final decisions to know the decisions that create a high-performance chip. Ren says the approach should halve the engineering effort required to produce the chip, while producing a chip that matches or exceeds the performance of what the human designed.
“You can design the chips more efficiently,” Ren says. “It also allows you to explore more design space, which can make you better at chips.”
Nvidia started making graphics cards for players, but quickly saw the same chips as potential machine learning algorithms, and is currently the leading creator of high-end AI chips. Ren says Nvidia plans to bring the AI-crafted chips to market, but declined to say how soon. In the more distant future, he says, “you’ll probably see the bulk of the chip designed with AI.”
Reinforcement learning was used by computers to train them to play complex games, including the Go board game, with superhuman skills, without explicit instructions about the rules of the game or the principles of good play. It shows promise various practical applications, including trains robots to capture new objects, aircraft aircraft, and algorithmic stock trading.
Song HanAccording to an assistant professor of electrical and computer engineering at MIT, learning reinforcement shows great potential for improving chip design, as with a game like Go, it can be difficult to predict good decisions without years of experience and practice.
His research team recently he developed a tool uses reinforcement learning to identify the optimal size of different transistors on a computer chip by analyzing different chip designs in the simulation. It is important to be able to transfer what you have learned from one type of chip to another, which promises to lower the cost of automating the process. In the experiments, the AI tool produced circuit designs that were 2.3 times more energy efficient, while creating one-fifth of the interference designed by human engineers. MIT researchers are working on AI algorithms, along with new chip designs, to take advantage of both.
Other players in the industry, especially those who invest in the development and use of AI, want to take AI as a tool for designing chips.
[ad_2]
Source link
