Has the Intel CPU out-of-stock issue really been resolved? Is it really back to the pinnacle of the industry?

According to statistics, in the history of the past 50 years, AMD will lag behind Intel by one or two generations in CPU technology most of the time.


Intel's 10nm process has been slow to produce energy, and the 14nm process has to be improved for three generations. It is still the main force of high-performance CPUs, which also gives AMD a chance to surpass. AMD relies on Ryzen and Xiaolong processors in the desktop, notebook and server markets to bring a lot of pressure to the friend Intel, especially after the launch of the 7nm Zen2 architecture processor this year, AMD is leading the performance and technology for the first time.


 



 


Intel has the most advanced CPU technology on the planet. Why is it slow to respond to AMD's competition? This question is probably something everyone wants to know. Intel’s senior vice president and chief engineer Murthy Renduchintala recently presented at the UBS TMT Technology Conference. Explained.


According to Murthy Renduchintala, one of the biggest challenges that Intel faces when developing native single-chip processors is not only integrating the logic core IP, but also considering the signal and analog IP issues, and connecting these IP units Up the unit circuit.


The problem does not stop there. These signal, analog, and connection units do not benefit from advanced technology, so 50% of Intel's one-year budget is spent on transplanting these technologies.


However, Murthy Renduchintala said that they have found a way to decouple the CPU architecture from the process technology, which can be completed at a different pace, which will increase the speed of product roadmap upgrades. In addition, he added that Intel will return to its leadership position.


In fact, the difficulties mentioned by Intel are also the difficulties of AMD. The IO part is very difficult to handle. AMD’s approach is to separate the two and let the CPU use the advanced 7nm process, and the IO core can use the 14/12nm process. Sooner or later, I will also go this way.


cache
Processed in 0.004164 Second.