The wonders of processor acronyms ¼: CPU

Nowadays, most people own a personal computer ( PC), smartphone or electronic devices embedding chipset, memory, processor etc..

Even if most might approximately know the basics about RAM, CPU, etc…, it is unlikely we know the difference between CPU, GPU, APU, SOC.

Let’s start with the CPU, maybe the most known component of a computer with the GPU.

The CPU, also called Processor, stands for Central Processing Unit. 

It is the head of your system, the chef orchestration doing most of the tasks and calculations. 

From the CPU 8086 at 5Mhz with 29 000 transistors engraved at 3000nm in 1978, the CPUs evolved to nowadays 3~4Ghz engraved at 7nm, passing by the PentiumIII or AMD Athlon at 1.5Ghz with 42 million transistors engraved at 180nm in 2000. 

Based on the famous Moore law, the founders have kept improving Frequency and number of transistors as it was the leverage to speed up your computer. 

To do this, they added more and more transistors and renewed their architecture to allow more instructions per cycle and improved the engraving fineness to minimize the traveled distance by electrons and also help to add more transistors on a smaller surface.

Eventually after years of development in those directions, processors being faster, one issue faced was the interconnection and intercommunication between the CPU and other dependencies such as the RAM due to the low speed of data transfer on the data highway also called BUSes between them.

About frequency, founders also faced a limit of reachable frequency due to, what was considered at that time, a limit of physics.

Indeed, the technologies and knowledge back in the days couldn’t face, as per exemple, the engraving fineness, which was thought to be limited at 70 nm due to the electron behaviour at that scale. 

But time told us that science can provide solutions and better knowledge of the matter if we give it time. Which allows us nowadays to get an engraving fineness of 10 to 7nm and to be even less according to Intel promising an “Angstrom era” revolution by 2024 with a fineness of 1.8nm. (ndlr: an Angstrom is typically the order of magnitude of an atom).

In order to keep improving the power of their CPU, other ways of improvement were the change of the 32bits architecture to 64 bits (increasing the size of data packets and so the number of instructions per cycle) as well as the duplication of CPUs working together,  the multi-core processors were born. 

It is now common to see CPUs with 8,16,32, or 64 cores.

And the new challenge now is to make them more and more eco-friendly by lowering their energy consumption. It is why, lately, we see cores dedicated per usage for low and high energy consumption depending on the tasks .

Along the years, 2 mastodons emerged: Intel and AMD.

PAGE TOP