当前位置：网站首页>Graphics card, GPU, CPU, CUDA, video memory, rtx/gtx and viewing mode

Graphics card, GPU, CPU, CUDA, video memory, rtx/gtx and viewing mode

2022-06-26 03:18:00 【MengYa_ Dream】

Learning resources ：

CPU and GPU What's the difference ？https://zhuanlan.zhihu.com/p/156171120

After reading, I can understand ：GPU Relationship with graphics card ,GPU and CUDA What does it matter ？https://blog.csdn.net/weixin_42306148/article/details/117226047?spm=1001.2101.3001.6650.1&utm_medium=distribute.pc_relevant.none-task-blog-2~default~CTRLIST~default-1-117226047-blog-52642366.pc_relevant_antiscanv4&depth_1-utm_source=distribute.pc_relevant.none-task-blog-2~default~CTRLIST~default-1-117226047-blog-52642366.pc_relevant_antiscanv4&utm_relevant_index=1

Catalog

1 Visual visual memory understanding - video ：CPU and GPU The difference between

2 The concept of graphics card

3 GPU The concept of

4 CPU The concept of

5 CUDA The concept of

5 CPU and GPU The relationship between

6 CUDA and GPU The relationship between

7 Concept of video memory

1 Visual visual memory understanding - video ：CPU and GPU The difference between

CPU： It can be understood that pigments are emitted successively , come together , An image that presents a smiling face , It takes a long time , Analogical CPU Calculation principle of , Step by step iterative calculation .

GPU： Multiple tubes of paint are emitted simultaneously , Draw a complete portrait , Time is spent in a moment , Analogical GPU Calculation principle of , Massive high parallel computing .

therefore , In experiments, we usually want to calculate in GPU In the middle of , Common code ：

# select device
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
print("device: %s" % device)

You can see that... Does not appear in the code GPU, actually , It suggests that GPU and CUDA There is a certain connection between , The following is a detailed study record .

2 The concept of graphics card

The graphics card （Video card,Graphics card） Full name: display interface card , Also called display adapter , Is the most basic configuration of the computer 、 One of the most important accessories . Just like computer networking requires a network card , To display the data in the host on the screen, you need a graphics card . therefore , A graphics card is a device used by a computer to convert digital to analog signals , Undertake the task of outputting display graphics . say concretely , The graphics card is connected to the main board of the computer , It converts the digital signal of the computer into an analog signal for display . The original graphics card is generally integrated on the motherboard , Only complete the most basic signal output work , You don't have to deal with data . With the rapid development of graphics cards , And that's what happened GPU The concept of , Graphics cards are also divided into Independent graphics card and integrated graphics . The biggest difference between independent graphics card and integrated graphics card is in performance and power consumption . The former is more powerful , The latter is limited by integrated circuit and cost , Only the performance of top-level core display can reach the performance of entry-level unique display ; The former has independent video memory, and the latter adopts the shared video memory design with memory , It also affects performance ; The former is more powerful in power consumption , The latter has lower power consumption .

Independent graphics card ： Strong performance , Independent memory , High power consumption .
integrated graphics ： Weak performance , Share video memory with memory , Low power consumption .

3 GPU The concept of

GPU The concept is made up of Nvidia The company 1999 Put forward in .GPU It's a chip on the graphics card , It's like CPU It's a chip on the motherboard . that 1999 Years ago, there was no... On the graphics card GPU Do you ？ Of course. , But no one named it at that time , It has not attracted enough attention , The development is relatively slow . since Nvidia Put forward GPU After this concept ,GPU Entered a period of rapid development . Simply speaking , It has gone through the following stages of development ：
① For graphics rendering only , This function is GPU The purpose of ,Graphic Processing Unit, Graphic processing unit ;
②GPU It's too wasteful to use only for graphics processing , It should be used to do more work , For example, floating point operations . How to do it? ？ Give the floating-point operation directly to GPU It can't be done , Because it can only be used for graphics processing （ At that time ）. The easiest thing to think of , Is to do some processing of floating-point operation , Wrap into graphics rendering tasks , And then to GPU To do it . This is it. GPGPU（General Purpose GPU） The concept of . But there is one drawback , You must have some knowledge of graphics , Otherwise you don't know how to pack . In order for people who do not understand graphics to experience GPU The power of computation ,Nvidia The company has put forward CUDA The concept of .

4 CPU The concept of

a central processor （central processing unit, abbreviation CPU） As the computing and control core of computer system , It's information processing 、 The final execution unit of the program .

5 CUDA The concept of

CUDA(Compute Unified Device Architecture), General parallel computing architecture , It's a computing platform . It contains CUDA Instruction set architecture and GPU Internal parallel computing engine . All you have to do is use something like C Linguistic CUDA C Language , Can be developed CUDA Program , Thus, it can be used more conveniently GPU Powerful computing power , Instead of wrapping the computing task into a graphics rendering task as before , Give it to GPU Handle . Be careful , Not all GPU All support CUDA.

5 CPU and GPU The relationship between

In the absence of GPU Before , Basically, all tasks are assigned to CPU To do the . Yes GPU after , There is a division of labor between the two , CPU Responsible for logical processing and serial computing ,GPU It focuses on performing highly threaded parallel processing tasks （ Large scale computing tasks ）. Why is there such a division of labor ？ This is determined by the hardware composition of the two . It can be seen that , CPU yes “ Lord （host）” and GPU yes “ from （device）”,GPU No matter how fast it develops , It can only be for CPU Share the work , Instead of replacing CPU.

6 CUDA and GPU The relationship between

CUDA yes GPU Implementation tools for ;GPU yes CUDA The hardware foundation of .

7 Concept of video memory

memory , Also called frame cache , Its function is to store the rendering data processed or to be extracted by the graphics card chip . Just like the memory of a computer , Video memory is a part used to store graphic information to be processed . It can also be understood as the memory of the graphics card , Computer view graphics card method ：16G memory