CPUs and GPUs, although used for different functions, have some things in common. One of these commonalities is that they are based on microscopic transistors . These transistors are measured in nanometers . We'll explain what this parameter is and what it's used for.
Table of Contents
What are nanometers
It is the unit of measure of length that is equal to one billionth of a meter.
Nanometers is expressed as nmElectronic chips are manufactured in what is called a foundry . Some of the most important foundries are TSMC, Intel, GlobalFoundries, and Samsung, among others. These companies measure their different manufacturing processes in nanometers.
This measurement was initially a standard for referring to transistor size . However, due to the physical limitations of manufacturing technology, this measurement system was lost , and now foundries express it in an approximate way . Currently, this measurement is more for marketing purposes than anything else.
Each foundry has its own manufacturing processes , making it very difficult to compare lithographs at the nanometer scale.
Now, nanometers refer to the chip manufacturing node . They aim to reflect the density of transistors that can be "printed" on a chip. The higher the density in a given space, the more powerful the manufactured chips will be. It simply reflects density.
👉 Which is the best processor on the market ? 👈
What are nanometers for?
Increased density of transistors
It determines the number of logic ports and circuits that can fit within a silicon chip of a specific size . The 14nm Core i9-9900K processor with a 174mm² die has almost 3 billion transistors. Ryzen 3.000 processors have approximately 3.9 billion transistors on a 74mm² die using a 7nm node.
Increase in frequencies
The higher the transistor density, the greater the number of states on a chip . This results in a higher number of instructions per cycle (IPC).
Improve energy efficiency
Reducing the size of the node requires less electron flow . The transistor changes state with less energy, improving energy efficiency . We can therefore do the same work with less energy, gaining performance per watt consumed.
Reduction of size
Miniaturizing the nodes allows for the integration of more transistors in increasingly smaller physical spaces . The silicon surface area can be reduced by increasing the number of transistors or by maintaining the same number. Another benefit of reducing the nanometer size is the reduction in heat generation.
Major chip foundries
This parameter is currently used by foundries to indicate the transistor density on a chip . It is expressed as an integer followed by the abbreviation nanometers. Intel currently offers 14nm processors , and TSMC manufactures AMD Ryzen processors using a 7nm lithography process.
There are currently four major chip manufacturers that use this parameter:
- Intel: Is world's largest chip maker and has its own foundries. The company designs and manufactures its own processors under its own nodes. It currently offers processors in 14nm and 10nm, in addition to already developing 7nm and 5nm lithographs. Does not manufacture chips for third parties, just for themselves.
- TSMC: The Taiwanese foundry is the one more has grown nowadays. Currently this foundry offers chips under the 7nm node and by 2021 it will start producing chips in 5nm. Between his Customers include AMD, Apple, Qualcomm, NVIDIA and even Intel. Interestingly, this Taiwanese foundry does not manufacture its own processors, only for third parties.
- Samsung: The South Korean company also has its own foundry for its chips and memories. Currently its most advanced node is the 8nm node, which is an optimized version of the 10nm node. Samsung manufactures own solutions and for third parties, like the GPUs of the new NVIDIA Ampere.
- Global Foundries: The American foundry with plants in several countries, has been left behind. At present only make chips on the 14nm and 12nm nodes, leaving the career of lithographs.
- SMIC: The Chinese foundry supported by the government works at full speed to compete with the rest of foundries. They are currently on the 22nm node and have the 14nm node practically ready. This foundry is involved in certain controversies, since has been sued on several occasions for alleged patent infringement.