After returning to the high-performance CPU market in 2017, AMD has put a lot of pressure on Intel in the desktop, notebook and server markets with its Ryzen and Dragon processors, especially after the 7nm Zen2 architecture this year, amds are leading both in performance and process for the first time. Compared with AMD, Intel’s side makes a lot of I meals unreadable.
14nm processor launched generation after generation, although this year, although the production of 10nm process, but mainly for low-power notebook, still no high-performance processor, this is not until next year to release, when AMD will launch 7nm EUV process Zen3 processor?
Intel has the most advanced CPU technology on earth, so why is it slow to respond to AMD competition? This is a question that everyone wants to know, explained murthy Renduchintala, Intel’s senior vice president and chief engineer, at UBS’s TMT technology conference.
According to Murthy Renduchintala, one of Intel’s biggest challenges in developing a native single-chip processor is not only to integrate logical core IP, but also to consider signal and analog IP issues, as well as the cell circuits that connect these IP units.
The problem doesn’t stop there, these signals, simulations, and connection units don’t benefit from advanced processes, so 50 percent of Intel’s annual budget is spent on porting these technologies.
From Murthy Renduchintala’s explanation, you should understand why Intel’s CPU architecture is combined with fixed processes, because Intel’s model is to customize different architectures for different processes, and the benefit is that each generation of processors can perform at its best. The disadvantage is that this is too inflexible, and once the process is delayed, it affects the entire processor upgrade plan.
Fortunately, Intel has solved this problem, and Murthy Renduchintala says they’ve found a way to delink the CPU architecture from the process and be able to do it at a different pace, so that the product roadmap upgrade is much faster.
But Murthy Renduchintala didn’t reveal details of how to delink and when to release the next generation of high-performance processors.
With Intel’s description, we should now understand why AMD in the 7nm Zen2 processor CPU core and IO core separation bar, Intel said the difficulty is also AMD difficulties, IO part is very difficult to deal with, AMD’s approach is to separate the two, let the CPU use advanced 7nm process, IO core using 14/12nm process can, and Intel sooner or later will go down this road.