Compared with traditional processors
The market has proven the benefits of Arm chips’ lower power consumption over the past 10 years. Following in the footsteps of Apple’s M chips, major technology companies such as Microsoft and Qualcomm are also actively investing in Arm architecture. AI PC. Microsoft’s Windows on Arm, Copilot+ PC plan, and Qualcomm’s Snapdragon X Elite processor are typical examples.
In addition to the manufacturing process
That most people are Brother Cell Phone List concerned about, Xu Dayong believes that there are two directions from which to observe the ripples generated by generative AI on semiconductors. “First, Moore’s Law has really almost come to an end, so the industry currently has some new practices to extend Moore’s Law, such as 2.5 and 3D packaging technologies. Subsequent production will not be able to cope with the large number of generative AI needs. It is worth observing. .
Another key battlefield is memory
Everyone is indeed working Estonia Phone Number List hard to reduce the number of parameters of large language models, but no matter what, the amount of memory required is still there. “So Near Memory and Compute In Memory (in-memory computing) are becoming areas that many people are investing in research. “Xu Dayong said, “Of course, if we get rid of pure theoretical discussions in the industry, heat dissipation will be a big focus, so the problem will return to how to find the best power consumption ratio.”Exclusive interview with Neuchips Chuangxin: How can Taiwanese startups seize the wave of generative AI.
