TSMC , the world’s leading semiconductor factory, confirmed that it will begin mass production of 3nm chips in the second half of 2022. According to the president of the Taiwanese company, Liu Deyin, the number of employees in the Tainan Science Park it will hit 20,000 when the time comes.
In theory, TSMC’s 3nm technology could enable Radeon GPUs and Ryzen CPUs to be made up to three times more complex and faster than current ones.
While TSMC promises at least 250 million transistors per square millimeter for its 3nm node, the reality is that it can approach 300 million. This means that by the end of 2022, TSMC will have the capacity to produce chips between 2.5 and 3 times denser than the 7nm technology currently used.
This density can be used to make existing complexity chips much smaller and cheaper, or allow for much more complex designs. For example AMD’s Navi 21 GPU, used in the Radeon 6800 and 6900, registers just under 27 billion transistors thanks to 7nm. In this way, a 3nm-based GPU of the same size could hold around 80 billion . A real madness.
However, AMD does not tend to use the latest TSMC production technology although the Taiwanese company is already producing chips for Apple iPhones in its new 5nm node. In fact, the red team has already revealed that Zen 4 will arrive in 2022 and will be based on TSMC’s 5nm.
Are Intel and Samsung left behind?
Meanwhile, Intel’s 7nm node was deferred to at least late 2022 and will likely not be seen until 2023 or later. It is estimated that the 7nm Intel will deliver around 200 to 250 million transistors per square millimeter. So it will sit somewhere between the 5nm and 7nm nodes of TSMC.
As for Samsung , the other big player in chip making, recently announced its plans to close the gap with TSMC with its 3nm in the second half of 2022.
Most importantly, however, the future looks very promising when it comes to PC performance, at least for the next four to five years.