What are the differences between a CPU and a GPU ?

Deepak Kumar
3 min readNov 21, 2018

--

While buying a laptop or PC, you must have heard of the terms CPU and GPU and if you’re a techie guy, you must be aware of the acronyms of GPU and CPU. If you’re not, GPU stands for Graphics Processing Unit and CPU stands for Central Processing Unit. Before coming to the main differences between them, let’s talk about the basic similarity.

A CPU and a GPU both are processors which read and execute program instructions. Both contains millions/billions of transistors which alter or manipulate signals as required. Both work on the same instruction sets of Logical Gates.

Now we know that both are made up of transistors and processes instructions like a silicon chip. Basics seems to be same in both the processing units but the actual picture is a bit different. Let’s start the comparison of both,

Basic works of both :-

CPU is responsible for logical processes, input/output (I/O) operations such as adding two numbers, Job scheduling, Graphic User Interface, printing etc. while GPU is responsible for rendering videos and displaying graphical output on monitor at higher refresh rates.

In other words, CPU has a good amount logical(conditional) tasks to do, anything from adding two numbers to managing all the connected hardware while GPU has just one kind of repetitive and parallel computing task such as real-time rendering videos and anything which need parallel computing like machine learning.

Architectures of both :-

Talking about architecture, CPU has several sub-units such as Arithmetic Logical Unit (ALU), Control Unit (CU), Cache Memory and Dynamic Random Access Memory (DRAM) while a GPU lack an ALU, and have of small cache memories and control unit for thousands of cores with a DRAM.

Source — NVIDIA, CC BY 3.0, via Wikimedia Commons

A CPU has few cores but powerful for making complex & logical operations, ranging from 1- 61 cores for example Xeon Phi has 61- cores which can only execute 61 complex instructions simultaneously during per clock cycle (in a second) while a GPU has thousands of cores that render thousands of pixels simultaneously during per clock cycle, and execute a complex 3-D video on your monitor. For example, Nvidia GTX 1080 has 2560 Shader cores, thus it can execute 2560 instructions simultaneously during one clock cycle.

Source — Mmanss

So, a CPU comprises of few to several of brainer cores and a GPU comprises of thousands of dumb cores, but being dumb is not bad because a GPU can be 250 times faster than a CPU. Thus, both has their own abilities to offer.

Conclusion

In general language , CPU is doing less jobs but complex ones while GPU is doing a hell lot of jobs but dumb ones.

Both CPU and GPU work on the same principle of processing data to make useful information (Logical and graphical both) but GPU is specialized to do parallel and repetitive smalls tasks at higher rates.

--

--

Deepak Kumar
Deepak Kumar

Written by Deepak Kumar

Graduate Frontend Engineer at Doorpix | Sometimes Code, Sometimes Write. On a journey to live a better life | https://dazzlerkumar.vercel.app/

No responses yet