Video & Live Streaming

What is the Graphics Processing Unit (GPU)? GPU vs CPU

Shayek Ahamed | June 27, 2019 | 506 views | 3 People said helpful.

What is Graphics Processing Unit (GPU)?

In computer terminology, the term GPU stands for Graphics Processing Unit. The Graphics Processing Unit (GPU) is a specially designed electronic circuit or processor which is primarily used to manage and accelerate the performance of the computer’s 2D/3D graphics, video, visualization, and display. Sometimes, it also called visual processing unit.

Features of a Graphics Processing Unit (GPU)

Graphics Processing Unit or GPU is the core and integral part of the Graphics Card, where all the graphics processing takes place. Here are some important features of a GPU, which is designed to perform video and graphics much faster without degrading the performance of the CPU (Central Processing Unit).

  • Digital output to flat panel display monitors
  • Rendering polygons
  • Application support for high-intensity graphics software such as AutoCAD and some sort of gaming applications
  • Texture mapping
  • 2D or 3D graphics
  • Hardware overlays
  • MPEG decoding
  • Video editing
  • Support for YUV color space
    high-quality gaming experiences
  • Video Encoding, Decoding or Transcoding
  • Video Scaling and Filtering
  • Advanced Motion Adaptive Deinterlacing
  • Video Color Correction
  • High-Definition MPEG-2 and WMV Hardware Acceleration
  • Machine Learning

All the above mentioned features are designed to reduce the work of the CPU and process faster video and graphics.

History of GPU in computing

In the very early stage of computing, the GPU was not included in the PCs, therefore, the Central Processing Unit or commonly known CPU had to perform all these mathematical calculations and graphics operations. As software demands increased, more graphics intensive applications (especially in video games and graphics designing) were introduced, the demands created pressure on the CPU and obviously degraded the performance. Therefore, the Graphics Processing Unit (GPU), a separate processor, was introduced as a way to disburden the CPU from graphical tasks.

First commercial GPU was introduced in the market by NVIDIA on August 31, 1999, which was called the GeForce 256. The GeForce 256 was able to process 10 million polygons in every single second, allowing it to disburden a significant amount of graphics processing from the CPU.

The success of the first GPU caused both hardware and software developers alike to quickly accept GPU support. Consequently, motherboards were developed with much faster PCI and AGP slots designed especially for the graphics card. Software APIs, for example, OpenGL and Direct3D were formed to aid developers to make use of GPUs in their applications. Today, we are experiencing, dedicated graphics processing which is standard for all platforms from desktop PCs, laptops, smartphones to video game consoles. Manufacturers, like Nvidia, AMD are dominating the market still now and according to a recent survey by Ranker.com, Nvidia is considered as the best GPU manufacturer in the world, followed by AMD. Intel, a world renowned manufacturer, is paying more attention about the graphics chip business, although, it currently has a very limited presence, recently announced it’s first “discrete” graphics chip (GPU) is coming in 2020.

What is the Central Processing Unit (CPU)?

The Central Processing Unit or CPU is the core part of any computing device, performing most of the processing inside a computing device. It handles basic instructions and allocates the more complicated tasks to other specific chips to perform properly. To control and manage instructions and data flow, the CPU depends greatly on a chipset, a group of microchips situated on the computer’s motherboard.

The very first CPU, the 4004 unit, was developed by Intel in the 1970s. At that time, most CPUs were designed with single “core”, that means, only one operation could be processed at a time. Later, due to massive improvements in computer chip design, research, and manufacturing, the market advanced to faster dual-core and multi-core CPUs which are now performing multiple operations at a time.

What are the major differences between CPU and GPU?

Modern CPUs have a great number of cores to perform various sorts of complex computing. However, the basic design and purposes have not really changed. They are mostly applicable for solving problems that need analyzing through or interpreting complex logic in codes. On the other hand, a Graphical Processing Unit or GPU, has smaller-sized but plenty of logical cores like. ALU or Arithmetic Logic Units, control units, and memory cache. Its basic design is to process a set of easy and more unique computations in parallel.

Graphics Processing Unit (GPU): CPU vs GPU
Figure: CPU vs GPU

Although GPU was certainly around since from the availability of gaming applications. But it was popularized when NVidia marketed Geforce 256 (world’s first GPU). At first, GPU was designed as dedicated graphical rendering subservient of different types of computer games. Later on, it was powered by other geometric calculations like rotating verticals into different coordinate systems or transforming polygons. Though, Geforce 256 was designed for basic gaming and computing needs, but, new cutting edge turning architecture has been introduced in its latest version GeForce GTX 1650.

GPU Vs CPU: GPU is not a substitute of CPU

Well, there is no existence of any written agreement that says GPU is always faster, rather the performance all depends on the applications you use. Basic level of differences between CPU and GPU could involve the comparison of cores and processing speeds. In term of CPU, the number of cores is limited and the processing speed is high, whereas, in GPU the number of cores is large and the processing speeds are comparatively slower.

Before we go further, we should consider some important terms such as Latency and Bandwidth, related to this topic. Generally, Latency is a (certain) time duration taken for an action and bandwidth is the number of processes that can be done in a certain time. CPU mainly focuses on the latency. We can agree to this by the fact that there is a high processing speed on the given limited number of cores. In contrast, GPU concentrates on the bandwidth. Since the same operation is processing in parallel, many computations (mathematical calculations) can be processed in a certain amount of time.

One more significant difference between CPU and GPU is that since both the CPU and GPU do data-parallel tasks, GPUs are designed in such a way that more transistors are involved to data processing rather than data caching and flow control. It accelerates the performance by allocating compute-intensive portions of the application to the GPU, while the remaining code still runs on the CPU. Therefore, from a user’s perspective, applications just run much faster, while in general-purpose computing is still the CPU’s domain, GPUs are the hardware backbone of nearly all intensive computational applications.

In Graphics Processing Unit (GPU) more semiconductors are involved in the data processing. GPU provides superior user experience because applications run faster than before. It directs compute intensive portions of the applications to GPU while other codes run on the CPU, thus it fast-tracks the performance.

Use cases of GPU

GPUs are intensively using in various fields of computation. Here is a list of popular use cases for GPU or Graphic Processing Unit.

  • Deep Machine Learning
  • Image Processing, Graphics Rendering
  • 3D Mapping
  • Games
  • Financial Analysis
  • Video Analysis
  • Video Encoding, Decoding, and Transcoding
  • Medical Imaging

What’s new in GPU and CPU?

Some terms related to this article to consider:

APU: The AMD Accelerated Processing Unit or APU, formerly known as Fusion, is a series of 64-bit microprocessors which has been designed to perform as a CPU and GPU on a single die.Ref: Wikipedia

TPU: A tensor processing unit (TPU) is an AI accelerator application-specific integrated circuit (ASIC) developed by Google specifically for neural network machine learning. Ref: Wikipedia

Are CPUs obsolete?/ Are CPUs out of date?

The Central Processing Unit is the main unit of a computer which performs a significant amount of the processing inside a computer device. Many tasks performed by PC operating systems and applications are still best fit to CPUs, and much work is needed to accelerate a program using a GPU. As we know, many existing software uses the x86 architecture, but GPUs require different programming techniques and it missed several important features required for operating systems. Therefore, a general transition from CPU to GPU is very difficult.

Conclusion

In conclusion, there are significant differences between the two computing terms ( CPU and GPU). CPU is best in sequential and GPU is for parallel computing. Therefore, GPU is not a replacement for CPU architecture, rather, they are powerful tools for the existing infrastructure.