AMD processors will integrate AI functions next year: beyond CPU/GPU

AMD processors will integrate AI functions next year: beyond CPU/GPU

AMD processors will integrate AI functions next year: beyond CPU/GPU.

AMD benefits from more than $ 53 billion acquisitions, AMD processors will integrate AI functions next year: beyond CPU/GPU. 

In February this year, AMD officially announced the completion of the acquisition of Xilinx.

The previous $ 35 billion acquisition has been worth more than $ 53 billion with AMD ’s stock price soaring .

This transaction was not in vain. , AMD CEO Lisa Sui said that next year’s processors will integrate AI capabilities.

Xilinx was the world’s largest FFPGA chip manufacturer before, and AMD ‘s acquisition of them was mainly because of the powerful capabilities of FPGA chips in the data center, especially in the field of AI , which are the key chips for high-performance computing.

It spent $ 16.7 billion to acquire Altera , the largest FPGA manufacturer at the time .

Although another competitor NVIDIA did not acquire FPGA chips, their GPUs are now the king in the AI ​​field. The newly released H100 accelerator card has achieved 32PFlops ( 320 million times) of AI calculation examples, which is the previous generation A100 accelerator card . 3 times the performance.

After the acquisition of Xilinx, when AMD launched a processor integrating FPGA has become the focus of everyone’s attention.

At today’s financial report meeting, AMD CEO Lisa Sui also confirmed that the company will launch a processor integrating AI capabilities in 2023 .

She said AMD has the industry’s best portfolio of high-performance and adaptive computing engines and has the opportunity to provide a stronger product portfolio with their expanded technology portfolio.

In addition to hardware, AMD is also developing unified software to help developers leverage AI capabilities for inference and training.

Lisa Sui said that Xilinx will enable AMD to have a broader product portfolio in the field of AI hardware, beyond the current CPU and GPU capabilities.