Graphics card knowledge book: easy to understand the knowledge of graphics card literacy and graphics card model parameters detailed

For home ordinary home, business office, audio and video entertainment, Internet classes, stock trading, online stores and other needs, for this type of demand for basic graphics performance requirements are not high, usually entry graphics card or core graphics can cope, but if it involves 3D games or 3D design and other needs, then naturally there are certain requirements for the performance of the graphics card, the choice of graphics card has undoubtedly become of paramount importance. Take games for example, if we play a 3D game, there is a low FPS frame rate, the frame rate is often below 60 fps, the screen card into PPT, most of this situation is due to the performance of the graphics card is not enough to cause. And consumers in the purchase of graphics cards, often because many white people do not understand the graphics card, in the purchase of graphics cards only look at the size of the memory, thus falling into the purchase of misunderstanding, which is the main reason for not understanding the parameters of the graphics card. This article, the installation of the home will share a graphics card knowledge of science and technology, in an easy-to-understand way to popularize the knowledge of graphics cards and graphics card model parameters detailed.

Graphics card knowledge book: easy to understand the knowledge of graphics card literacy and graphics card model parameters detailed

First of all, let's understand the role of the graphics card, so in the computer, what role does the graphics card play?

The graphics card, also known as the display card, is an important part of the computer, the graphics card is mainly responsible for the output display graphics tasks. The main role of the graphics card is to process the instructions and data provided by the CPU into text or images that the monitor can accept and then display them, a process that usually includes four steps, as follows.

1. firstly, the CPU transfers the data through the bus to the display chip of the graphics card.

2. the display chip processes the data and stores the processed data temporarily in the video memory.

3. The video memory transfers the data to the RAMDAC (random digital-to-analog converter memory) and performs the digital-to-analog conversion;

4. RAMDAC outputs the analog high frequency signal to the monitor imaging through the display interface.

If this expression is still beyond your comprehension, let's take an easy-to-understand approach to make you understand the role of the graphics card.

For example, suppose the CPU is equivalent to the human brain, here we compare the graphics card to a painter, the CPU (equivalent to the brain idea) first needs to tell the graphics card what needs to be painted, the graphics card will follow the CPU's requirements to paint a picture, a picture is equivalent to the number of frames, then the monitor will be a coherent high-speed playback of these pictures, we can see a coherent monitor screen picture on the monitor screen. The better the performance of the graphics card (the famous painter), the more and better pictures can be drawn, so that we can see a more smooth and clear picture through the monitor.

The second chapter: the difference between nuclear and independent display and the two camps of independent display

Whether desktop or notebook, graphics cards are mainly divided into two types, respectively, core graphics and discrete graphics, which can also be called integrated graphics, but the early integrated graphics are integrated in the motherboard, and now the integrated graphics are built into the CPU, which means that the CPU is equipped with a display chip inside, we now call the core graphics or core graphics, such as intel CPU model without F suffix or AMD CPU model suffix with G processor are built-in core graphics card. But due to CPU power control, heat problems, so the CPU built-in core graphics performance are relatively low-end, even the strongest performance core graphics can only be equivalent to the entry-level low-end level of discrete graphics cards.

The independent graphics card, as can be seen from the name, it is an independent hardware equipped with a display chip, it is not like a core graphics card, you need to divide some capacity from the memory to the memory, but in the independent graphics card has been equipped with its own memory particles, because it is independent hardware, it also has its own exclusive graphics card cooler, in general, the stronger the performance of the independent graphics card, the better the manufacturer's factory-configured cooler to better help the graphics card GPU chip heat dissipation, to meet the conditions of better heat dissipation, then the graphics card performance can be stronger, if you have high requirements for graphics performance, only consider the independent graphics card.

Independent graphics cards, currently divided into two camps, namely NVIDIA and AMD, the two manufacturers of graphics chips, which we often say N card and A card, N card in the largest market share, from entry to the flagship product line is more abundant, and in the high-end graphics market is far ahead, relatively more advantageous in technology, power consumption is also lower than the A card, optimization is also better, while the A card product line is not particularly comprehensive, less user-selective, but A card is relatively cost-effective, the same level of graphics cards, the price is cheaper and more affordable.

Graphics card knowledge book: easy to understand the knowledge of graphics card literacy and graphics card model parameters detailed

Regarding the non-public and public versions of graphics cards, NVIDIA and AMD's own graphics cards are called public versions. The public version refers to the original design (or authorized design) of the PCB version by NVIDIA or AMD at the beginning of the new release, including circuitry, interfaces, materials, and heat dissipation are all unified standards, and the core frequencies of the graphics cards are also consistent, and specify the graphics cards produced by OEMs.

Graphics card knowledge book: easy to understand the knowledge of graphics card literacy and graphics card model parameters detailed

Public Edition Graphics Cards

However, the majority of graphics cards we buy in the market are non-public versions, the market ASUS, MSI, Gigabyte, ShadowChip, Seven Rainbow, Kongsheng, Sooty, Yingzong, MaxSun, Sapphire, Dylan and other brands are OEMs, using the GPU chips provided by NVIDIA and AMD, which we call "non-public". In the case of the same GPU chip graphics card, these OEMs will design different graphics card appearance, workmanship, materials, interfaces, lighting, heat dissipation, and divided into low-end, mid-range, high-end versions sold at different prices, including some models of graphics cards in the core frequency preset higher, of course, the performance difference is not too big, or the main difference between workmanship and materials and heat dissipation.

Graphics card knowledge book: easy to understand the knowledge of graphics card literacy and graphics card model parameters detailed

Non-public version graphics cards

Chapter 2: Independent graphics card model naming rules knowledge

1, the N card chip model meaning.

N card models are mainly prefix + digital composition, the current N card model prefix were GT, GTX, RTX three forms of the beginning of the number behind the model is the greater the performance!

The GT starts with a low-end entry position, such as the GT1030.

The beginning of GTX is positioned in the low-end or mid-range mainstream, such as GTX1650, GTX1660Super, GTX1660Ti.

RTX begins with a model positioned mid-range or above, and RTX begins with an indication that it supports ray tracing technology, such as RTX3060, RTX3060Ti, RTX3090.

Ti ending: enhanced version.

Super ending: Super is somewhat similar to Ti, which you can understand as a super version.

Graphics card knowledge book: easy to understand the knowledge of graphics card literacy and graphics card model parameters detailed

2、The meaning of the chip type of A card.

A card model is mainly composed of prefix + number, the current A card model prefix is currently the beginning of the RX, and the number model behind it is the more high-performance the stronger!

XT Ending: Enhanced version.

Graphics card knowledge book: easy to understand the knowledge of graphics card literacy and graphics card model parameters detailed

The third chapter section: the core parameters of the graphics card popular knowledge

1、Graphics card architecture

The architecture of the graphics card is the architecture, which is the specification provided by the graphics card manufacturer for the products belonging to the same series of graphics cards, generally different brands (NVIDIA and AMD) or different generations, the architecture of the product is also different. nvidia and AMD will continue to launch a new generation of graphics chips, the architecture will also be improved and upgraded, generally speaking, the newer the graphics architecture the better the performance. As for the architecture of the graphics card, we can understand that the new car engine, based on the old car engine to improve the upgrade, so that the new engine performance is more superior and more fuel efficient, so the improvement and upgrade of the architecture of the graphics card performance has a huge impact, the same graphics card parameters, the new architecture may kill the performance of the old architecture.

2、Processing technology

Process process refers to the integrated circuit fineness when manufacturing graphics chips, the more advanced the process, it can reduce the size of the transistors, the same area of the wafer will be able to integrate more transistors, thus improving performance, while effectively reducing the processor power consumption and heat generation, and further upgrades in the architecture. For example, 22nm, 12nm, 7nm (nm), generally speaking the smaller the number represents the better the manufacturing precision.

3、Stream processor

The CUDA core we see in the parameters is the so-called stream processor (rendering tube), or SP for short. Theoretically, the higher the number of stream processors represents higher performance, and the more capable the graphics card is of drawing pictures, the faster it is. If you compare the number of stream processors of two graphics cards, only under the same architecture comparison is meaningful, different architecture down to compare stream processors is no reference value at all.

4、Core frequency

The core frequency of the graphics card refers to the operating frequency of the display core, its operating frequency can, to a certain extent, reflect the performance of the display core, the higher the core frequency, then the stronger the performance of the graphics card, of course, will cause the higher the power consumption of the graphics card. Graphics card overclocking is to improve performance by increasing the core frequency and memory frequency to achieve the purpose of improving performance, but should not exceed too much, otherwise there are unstable problems such as flickering screen.

5、Video memory type

Video memory, also known as frame cache, is used to store rendering data that has been processed or will be extracted by the graphics chip. Video memory is not an important factor in determining the performance of a graphics card, but it is definitely a key plus. The common types of graphics memory are GDDR6X, GDDR6, GDDR5X, GDDR5, and HBM, HBM2. A graphics card equipped with good memory will be better in terms of memory bandwidth, equivalent frequency and energy efficiency ratio, which is more conducive to improving the performance of the graphics card.

6、Display memory bandwidth

Memory bandwidth refers to the data transfer rate between the display chip and the memory, it is measured in bytes/second, memory bandwidth is one of the most important factors in determining the performance and speed of a graphics card, the formula for calculating the bandwidth of a graphics card is memory bit width x memory frequency = memory bandwidth. If the memory bit width is compared to a road, the wider the road, the more cars can pass, the memory frequency is equivalent to the speed of the car, the faster the memory frequency, the faster the data transfer speed, so the memory bit width and memory frequency of the graphics card has a greater impact on performance.

7、Display memory capacity.

The larger the memory capacity, the more data is cached and the stronger the graphics rendering capability. When we play a 3D game, the entire map and each element in the opening game the entire move into the video memory, if the image quality to the highest, so that the map of grass, trees, houses, vehicles more details, the amount of data, of course, more massive, need to be larger video memory support.

Of course, the size of the graphics card memory capacity does not determine the performance of a graphics card, memory is less important than the architecture, stream processor, core frequency, memory bandwidth, many people buy graphics cards only look at a few G capacity, which is one of the biggest misconceptions in the purchase of graphics cards, take the RTX3060 12G and RTX3060Ti 8G, performance is undoubtedly RTX3060Ti 8G higher, which also shows This also shows that the memory size does not represent the performance of a graphics card. If we compare the memory capacity to a parking lot, assuming that the road in the parking lot is not wide enough (bit width) and the cars are not fast enough (memory frequency), then the throughput of the parking lot will be very small, and this parking lot is completely wasted even if it is large (memory), which is like, if you pair an entry-level GT710 with 8G memory, it will still be the entry-level performance of the GT710, and it cannot become the RTX3060Ti .

8、Display interface knowledge

Common graphics card display interfaces are VGA, DVI, HDMI, DP, good and bad order is DP>HDMI>DVI>VGA.

Graphics card knowledge book: easy to understand the knowledge of graphics card literacy and graphics card model parameters detailed

The VGA (D-Sub) interface belongs to the analog signal, DP, HDMI, DVI belongs to the digital signal, of which DVI specifications are the lowest, DVI dual-channel support up to 1080P 120Hz and 2560 x 1600/60Hz. and does not support the transmission of audio signals, while HDMI, DP can support the transmission of audio signals, but HDMI and DP sub-versions However, HDMI and DP are divided into versions, different versions of HDMI and DP interfaces in different resolutions support different refresh rates, as follows.

HDMI 1.4 has a maximum bandwidth of 10.2Gdps and theoretically supports 1080p 144hz (up to 120Hz for most monitors), 2K 75hz, 3840 x 2160/30hz, 4096 x 2160/24hz.

The maximum bandwidth of HDMI 2.0 is 18Gbps, which theoretically supports 1080P 240hz, 2K 144hz, 4K 60hz, 5K 30hz.

The maximum bandwidth of HDMI 2.1 is 48Gbps, which theoretically supports 2K 240hz, 4K 144hz, 5K 60hz, and 8K 30hz.

DP1.2 version bandwidth up to 21.6Gbit/s, theoretical support 1080P 240Hz, 2K 144hz, 4K 60hz, 8K 30hz.

DP1.4 version bandwidth also reached 32.4Gbps, adding DSC display compression stream technology, thus theoretically supporting 1080P 240Hz, 2K 240Hz, 240Hz at 4K, 60Hz at 8K.

Many users use HDMI, but the display only supports a maximum refresh rate of 120Hz at 1080P, indicating that it is HDMI version 1.4, if you want to achieve HDMI 2.1, not a change of cable can be done, you need to monitor and graphics card at the same time both support HDMI 2.1 to be able to, for example, the current RTX30 series graphics cards support HDMI 2.1, for high Refresh rate of the gaming display, do not know what version of DP or HDMI, brainless DP line on the line.

9, ray tracing and DLSS technology

NVIDIA graphics cards (N cards) as long as the model beginning with RTX, that is, support with real-time ray tracing technology and DLSS, where real-time ray tracing technology can bring gamers more realistic game graphics, so that the game has a cinematic image quality, in fact, the so-called "ray tracing" technology, is the use of algorithms to simulate the physical characteristics of light in the real world, to achieve physically accurate shadows, reflections and refraction and global lighting, in the virtual game scene, so that the objects in the game more realistic sense.

Graphics card knowledge book: easy to understand the knowledge of graphics card literacy and graphics card model parameters detailed

The DLSS is called "Deep Learning Super Sampling" in Chinese, we understand that if your graphics card performance is not enough to support the smooth operation of high resolution, we can use DLSS technology technology to achieve the use of low-resolution hardware requirements, to achieve a display effect close to the high resolution. That is FidelityFX Super Resolution (FSR) technology, called "oversampling technology" in Chinese.

For example, for example, you have a host with RTX3060 graphics card, there is a 2K monitor, in a large game in 2K resolution can reach 60 fps, but if you have a 4K monitor, want to meet the native 4K resolution needs, then you may be in the settings to reduce the medium quality to maintain smooth, or use 2k resolution (or 4k resolution x 66% resolution), although you can make the game smooth, but in the 4K monitor display blurred, then it is certainly unpleasant.

And when the performance of the graphics card is not enough to support the smooth operation of high resolution, we can use DLSS technology to achieve the meaning of the existence of DLSS or the real use of the scenario is.

Note: The following data is just an example to help you understand better.

1. hardware consumption using 2k resolution or (4k resolution x 66% resolution).

2. Realize the display effect close to the native 4k resolution. (For example, the display effect is improved from 4k*66% resolution, to 4k*80% resolution)

In other words, the hardware requirements of low resolution to achieve a display effect close to the high resolution, where the said close is not comparable or equivalent effect.

10、Multi-card interconnection

The so-called multi-card interconnect is actually a technology that allows two or more discrete graphics cards to be used simultaneously to increase GPU computing power. Simply put, having two graphics cards in one computer at the same time will undoubtedly improve the graphics performance compared to one card. Currently, discrete graphics cards are mainly divided into two camps, NVIDIA and AMD, whether you choose NVIDIA or AMD have their own multi-card interconnection technology, NVIDIA graphics cards use SLI technology, while AMD graphics cards use CrossFire crossfire technology.

Graphics card knowledge book: easy to understand the knowledge of graphics card literacy and graphics card model parameters detailed

Generally speaking, as the performance of graphics cards is now higher and higher, few users consider multi-card interconnection, in addition to the strongest graphics card is currently unable to meet the performance of the time to consider multi-card interconnection. For example, dual graphics card interconnect, not double boost, the theory is to improve the performance of about 1.5 times, that is, we spend the price of two graphics cards, can only enjoy the performance of a half graphics card, obviously a bit uneconomical.

11、Graphics card brand

There are many brands of graphics cards, such as Asus, MSI, Gigabyte, Seven Rainbow, Shadowcharger, Kongsheng, Sotai, Image/AX Gaming Rebel, Onda, MaxSun, Yingtong, etc., as well as Sapphire and Dylan, which mainly focus on A cards, and Leadtek, which mainly focus on graphics professional cards. Graphics cards can not say which brand is the best, because the same display chip graphics cards, each brand has a low to high version, if the difference in workmanship materials, heat dissipation, etc..

What is the Beggar's Edition graphics card?

The word "beggar" is not a version defined by the brand manufacturer, but a version defined by the consumer. We know that for each graphics card brand, will be the same graphics card display chip into different versions, the cheapest version is called "beggar version", beggar version of the graphics card and the flagship graphics card gap in addition to the value, there may be differences in workmanship materials, heat sink, core frequency, RGB lighting, memory particles, display interface, etc., but because The difference between the flagship graphics card and the flagship graphics card may be in the workmanship, materials, cooler, core frequency, RGB lighting, memory particles, display interface, etc., but because the display chip is the same, there is basically no big difference in performance. For example, the flagship graphics card may be equipped with a metal backplane, RGB lighting, better workmanship and materials, better heat sink, better memory particles for overclocking, or the factory preset core frequency is higher (performance will be slightly improved), or the appearance may be more cool, these are the costs, the same display chip of the flagship version of the price is higher, but in fact, choose The beggar version is also a cost-effective choice, the current beggar version of the workmanship and materials is actually not bad, about the performance of the basic no difference.

What is a mining card?

The so-called mining card is a graphics card for mining, the user uses the computer to download the mining software and then run a specific algorithm, and the server communication can get the corresponding virtual coins, these virtual coins are also like the stock market, there are up and down quotes every day, according to the price of the day's coins can be sold, earn revenue! However, the state is currently cracking down on mining and virtual coin trading, very supportive! Mining mostly uses burning graphics cards to work, that is, 7 * 24 hours non-stop full load operation, generally a mining machine will be equipped with 6, 8 or more number of mid-range or high-end graphics cards, so the power consumption is huge, a serious waste of national power resources.

As mining are long-term 24-hour full load work, the GPU chip even if there is no problem, but the PCB and electronic components are accelerated aging, long-term full load temperature more affect the life of the components, so the purchase of second-hand graphics card friends must pay attention.

Is it necessary to choose a discrete graphics card for general office? When will I use a discrete graphics card?

If the family is a common household, business office, audio and video entertainment, Internet classes, speculation, online store needs, the graphics performance requirements are not high, is not necessary on the discrete graphics card, you can save some budget, after all, DIY computer is to match the hardware on demand to enhance the better cost performance. For example, the demand for medium/large 3D games, modeling and design, GPU rendering and other needs before you buy discrete graphics cards, discrete graphics cards are also divided into different models and different levels, which need to be matched as needed.

Is it okay to upgrade only the graphics card of an old host?

If the CPU performance is not weak, it is perfectly fine, although it does not exclude that the full performance of the graphics card is not possible because the CPU performance is too low. Some times, in the game, it is clear that the graphics card performance is completely competent, but still card, there is a game frame rate is very unstable or frame rate can not reach the standard of the graphics card, may be because of the CPU performance is low resulting in a short board, the barrel effect.

The installation budget is limited, choose high U low graphics or high graphics low U?

When the budget is ample, balanced with is undoubtedly the best program, the so-called balanced with, CPU and graphics card basically belongs to a class, such as mid-range mainstream CPU + mid-range mainstream or upper mid-range graphics card, hardware balanced with although good, but in the actual demand may be wasted, for example, I do not have high requirements for graphics cards, usually do not play games masterpiece, there is no graphics performance of office needs, the unique display may be superfluous, play a few years, the graphics card fan did not turn.

The so-called "high U low graphics" refers to a computer with a high CPU performance and a slightly lower graphics performance. The "high video and low U" refers to a computer with a strong graphics card but a weak CPU.

High U low graphics program is generally suitable for rendering class office, graphic design, multi-tasking program multi-purpose needs, servers, business office, video editing, ordinary household and other needs, choose entry-level graphics card or core graphics, can effectively reduce the cost of installation.

High display low U program, generally suitable for pure GPU computing, mining, most of the game, why is most of the game, not all games, because some games are eating CPU, some games are eating graphics, and some games are eating CPU and eating graphics, but most of the 3D games are mainly focused on the graphics card, the stronger the performance of the graphics card side screen performance will be better, FPS frame rate at the same time also higher. As long as the CPU to ensure that you can meet the needs of the game, high graphics low U is also a good choice program.

Whether it is high U and low graphics or high graphics and low U, no doubt to save the installed budget, but must be considered in the context of meeting their own needs.

How to see the high and low performance of two graphics cards?

Perhaps we are comparing the performance of two graphics cards, often due to different architectures or different generations, it is impossible to compare which one is stronger, for the white, the best and most intuitive way is to look at the graphics ladder chart.

The above is the installation of the home to share easy to understand the knowledge of graphics card literacy and graphics card model parameters detailed, this should be the most complete graphics card knowledge science and technology, of course, may be because of the length of the problem, there may also be no knowledge of graphics card science, later will continue to update in this article, I hope this article can help you.