The GPU or Graphic Processing Unit is a very important component in today’s computers since allows incredible graphics to be reproduced and, of course, Users can enjoy them. Hence, if you are a big fan of video games and high definition content you are interested in knowing everything about it.
This component works in a way similar to a CPU , but mainly analyzing computer images without considering other more general aspects.
In addition, there are several types, each with its pros and cons , and today you will also know them to get the graphics card that suits you best. We will also show you the differences between this and the CPU of a computer. Finally, we will discuss the extended concept of GPU acceleration .
Index:
What is the GPU or Graphic Processing Unit and what is it for in a computer?
This is a video component of technological devices (computers, teles, smartphones …) that has the function of processing the graphics . In addition, it helps to send all this information to the monitor.
This is, on the one hand, interprets the data that the CPU sends you, orders them and calculates the values that are of interest to you to store everything in your own memory; on the other, performs a processing of what is stored and transforms them , so that the output device in question (usually a monitor) can analyze and interpret it analogically.
It is worth mentioning that it is currently including other functionalities derived from supercomputing and network management as well as TV tuning, video capture, a special conetivity with peripherals …
In any case, his work includes the generation of anti-scaling, rendering, texture mapping, blurring, shading, tiling, speculation, etc.
This component, as we have started by saying, is very similar to a CPU, something like this version for the graphical context of the equipment . The graphical processing unit is a coprocessor that analyzes the images to try to lighten the load of a computer’s CPU .
Thus, this load is lighter when it is processed by the central unit and, therefore, can perform its functions more freely, without having to force it.
Physically speaking, we find with a hardware element, called a graphics card , that includes the set of softwares and other elements that work in these tasks. One of them is the GPU although it is already common to use both concepts as synonyms. After all, the processing unit is the main part of what is the card that, we repeat, is the physical element, the hardware that contains everything.
Usually, this is attached to the motherboard or motherboard, as is done with some of the other components of a computer. It can also be only a software and integrated chip on the board or even on the CPU, not attachable, that is, it is not a piece to add (we will see it later when we know the types of cards ).
Some ways to name this unit, in addition to a graphics card (which we have already seen are not the same but we have decided to give them the same meaning) and its graphics processing unit they are video card, video card or graphics accelerator.
Speaking of the complete concept of graphic or video card, we can not fail to mention that other parts are the GRAM, which is its graphic memory, with random access (RAMDAC), responsible for transforming a digital signal into analog, and the outputs that can be DVI, VGS, HDMI, DP …
As for its appearance, it occurs in the 80s and comes, like many other developments in the sector, from the hand of IBM, which used an Apple computer as an inspiration to achieve development of what would be the first video card. As a curious fact, indicate that this was interchangeable. It was 4 kB and was connected to a monitor.
GPU operation and features on a PC
Graphic processing units (GPUs) are designed with many processing cores that work at low clock frequencies . This is the opposite procedure to CPUs, which have few cores but very high clock frequencies.
The computer performs an analysis of the information that the graphic has sent and transforms it into information that the output device or peripheral can interpret.
Currently, the design of the graphics processing unit includes two main functions: pixel processing and vertices processing .
Next, we will see how each one works:
Vertex processing
This is a fairly simple job for the most modern graphics processing units since resource consumption is minimal.
The procedure is as follows: get the information from the vertices, which has already been calculated by the CPU and then process the rotation, the spatial order or segment of the vertex that is most visible at the graphic level. With this procedure you can continue with the pixelated.
The processing of vertices is done in advance by the central processor so that the space is rearranged, the graph is visible and the image can begin to be pixelated.
Pixel processing
This is when the graphics that the user can see are processed.
This is a fairly complex process and, in addition, requires a greater processing load . This occurs because all the layers and effects that are needed for textures that are much more worked and quite realistic are going to be applied.
Finally, once the graphics information is processed in the ways we have just seen, it is taken to a monitor, which will know how to interpret it.
Types of GPUs that exist today
There are currently three types of graphs in the market that differ mainly in the way they are implemented.
Let’s see what the main differences between them are:
Dedicated cards
These are the graphic units that have the greatest power . In addition, they have a series of features and specifications for which they are specifically prepared . Hence, they are also the most efficient. The fundamental characteristic of a dedicated card is that has a completely independent RAM , which will only be used by the graphics unit.
A dedicated card is one that is integrated in the motherboard connected by a specific port, although this is not always the most operational.
Integrated Graphics
Integrated units use system memory to perform their own functions. In the restrained computers are the most used to consume less. They are also used on tablets and smartphones.
The core of these cards used to always be on the motherboard ; but for the most recent models of AMD and Intel, they have decided to integrate these cards into the processor itself . This has been called the AMD Accelerated Processing Unit (for the AMD brand) and Intel HD Graphics (obviously for the Intel brand).
Hybrid cards
These types of cards are what we could call low cost. This is because they maintain very low prices but at the same time ensure certain levels of power suitable for virtually all home and office equipment.
Hybrid cards share system memory, although they integrate a small amount of their own memory that can be used for immediate tasks.
In addition to this type of cards, the latency time of the system memory is reduced.
What is that acceleration by GPU?
GPU acceleration has been used for quite some time although it is now that this feature is being used for web browsers, which is where we see it implemented more often. With this acceleration you get a much better performance and lower energy expenditure .
The accelerated coding by the graphics processing unit makes it possible to enjoy a image conversion 2 to 5 times greater when performing, whether redundancy, a conversion of graphics or videos . Therefore, the combined power of CPU plus GPU is used to accelerate the operation of those analysis applications and engineering.
GPU-accelerated computing causes to assign the work of the part of the applications where you need to perform calculations more comprehensively , while the lighter code is executed in CPU This will make the user believe that applications are running much faster.
In fact, this feature is used in the field of artificial intelligence (with novelties in drones and robots) and in certain applications that require high CPU consumption when solving complicated operations.
What are the differences between the CPU and a GPU?
We have talked about both throughout the text, but how are they different? This lies mainly in the architecture of both.
Although both elements are designed to process data, the GPU is more efficient for the calculation of graphical information , it has been created specifically for this. It is specialized for this task and may be less efficient in others.
The CPU, being a general purpose processor, can do all kinds of calculation , since it has been designed to process the data in series, being able, in fact, also to work the graphic information .
In fact, the graph takes work (graphic) from the CPU and when a certain level of acceleration is applied it can perform parallel processes in terms of graphics work.
The GPU is capable of running many more programs at the same time (in parallel), although these must be much more specific . The ability to perform parallel work that has the graph can multiply even a hundred percent the performance that a CPU will have when performing similar operations.
As for the physical design of a GPU, it has a greater amount of silicon, which allows it to perform very specific tasks efficiently. The CPU follows a Von Newman architecture and the GPU uses the Circulating Model, focused on parallel processing.