GPU或图形处理单元(Graphical Processing Unit)是一种专用微芯片，旨在以比CPU或中央处理单元(Central Processing Unit)快得多的速度专门执行与图形相关的任务。
一般来说，GPU(GPUs)处理视频游戏的视觉方面以及视频编辑软件、图形设计、3D 设计和其他类似工具。最近，GPU(GPUs)也被用于执行某些非图形任务，例如加密货币挖掘，这与其独特的内部设计配合得很好。
GPU和CPU之间的区别(Difference Between)
每台计算机，无论是笔记本电脑、平板电脑还是游戏机，都有一个CPU。CPU是机器的大脑。当您在设备上运行软件时，CPU会按照说明执行所有必要的数学运算。当然，每个微芯片都执行指令，那么CPU有什么特别之处呢？
大多数现代 CPU 都有四个CPU 内核(CPU cores)，尽管这个数字正在迅速增长。CPU可以同时执行的计算数量受内核数量的限制。但是CPU非常擅长处理任何类型的少量复杂计算，而且速度非常快，一个接一个。
然而，现代GPU(GPUs)的设计理念却截然不同。它们不是少数复杂的全功能处理内核，而是具有数百或数千个简单的处理器内核，旨在并行执行指令，特别是与图形相关的指令。
例如，以某种方式对图像中的像素进行着色的指令。现代(Modern)计算机显示器具有数百万像素。要以流畅的帧速率渲染和制作逼真的 3D 图形，需要这种强大的并行处理能力。所以如果你想同时处理很多很多简单的命令，你需要一个GPU。
GPU(Are GPUs)和显卡(Graphics Cards)是一回事(Same Thing)吗？
互换使用术语“ GPU(GPU) ”和“图形卡”已变得很普遍。这(Which)导致有人会说他们的计算机没有GPU，但他们真正的意思是计算机没有显卡。
什么是GPU？GPU特指微处理器本身，但GPU当然不能单独(GPU)运行。它需要其他组件来完成它的工作。因此，让我们回顾一下GPU(GPUs)的不同封装方式。
显卡
显卡是一种独立设备，可插入计算机主板上的标准插槽。它由印刷电路板、GPU、内存、显示连接器、电源连接器和GPU所需的其他各种组件组成。
显卡还具有专用的冷却系统，有助于将GPU保持在安全的工作温度内。显卡的主要好处是可以轻松更换新型号，而无需更换整个计算机或主板。
集成 GPU
集成GPU与(GPUs)CPU内置在同一外壳中。因此，配备其中之一的计算机不需要单独的图形卡即可工作。显示(Display)连接器由主板提供，GPU与(GPU)CPU共享相同的内存池。当然，CPU和GPU也共享相同的冷却系统和电源。
如今，集成GPU(GPUs)非常强大，有些甚至可以用于普通的视频游戏图形，但它们的主要优势是节省成本和空间。这(Which)就是为什么它们是许多笔记本电脑中流行的解决方案。
平板电脑和智能手机也集成了GPU(GPUs)，但这些是“片上系统”封装的一部分。在计算机上，只有CPU和GPU共享同一个封装。在SoC(SoCs)的情况下，构成整个系统的RAM、存储、外围控制器和其他组件都在同一个硅封装上。
专用 GPU
专用GPU不同于显卡和集成GPU(GPUs)。最容易将其视为计算机中内置的显卡。笔记本电脑是使用专用GPU(GPUs)的主要设备。
就像显卡一样，GPU有自己的内存和冷却系统，但与显卡不同的是，您无法更改或升级您的专用GPU。有少数笔记本电脑使用可互换的GPU模块，但这是一个罕见的例外。
外接显卡
最近的一项发展是外部显卡，它利用现代计算机上的高速端口。USB C(Thunderbolt 3 over USB C)上的 Thunderbolt 3 是首选解决方案。外部卡可以是集成解决方案，也可以只是一个外壳，您可以将其与任何适合的显卡一起使用。
对于在办公桌前需要更强大的GPU解决方案的笔记本电脑用户来说，这是一个很好的解决方案。(GPU)您可以在我们关于它们的文章中阅读更多关于外部 GPU的信息。(external GPUs)
了解 GPU 规格
如果您想知道如何在不同的显卡之间进行选择，请查看我们的显卡购买指南，了解(graphics card buyer’s guide)GPU规格的详细说明。如果您只想快速了解一下，以下是涉及GPU的关键数字：
- 处理器数量
- 内存量
- 总内存带宽
- 特殊功能，例如硬件光线追踪
不幸的是，知道这些数字或细节并不能告诉您给定GPU或显卡的性能如何。最后，最重要的是您要运行的应用程序的性能，您可以在在线发布的基准测试中轻松找到。
通用 GPU 计算任务
关于现代GPU(GPUs)要了解的最后一件重要的事情是它们可以用于某些非图形任务。由于GPU(GPUs)本质上由大量简单的处理器组成，因此任何需要并行完成大量计算的工作都可能受益于在GPU上运行。
例如，Nvidia品牌的卡使用CUDA（计算统一设备架构(Compute Unified Device Architecture)）指令。您可以使用CUDA(CUDA)编写软件，以便它利用 GPU 的独特设计来加速某些类型的计算。
还有OpenCL，它不附属于特定品牌的卡。GPU(GPUs)正被应用于大数据分析、机器学习、深度学习、加密货币挖掘，以及越来越多的过去需要超级计算机的高级工作。
我现在可以清楚地看到
这些是您需要了解的关于GPU(GPUs)是什么的最重要的事实。现在您知道它们是什么、它们的用途以及它们与其他类型的处理器有何不同。我们都应该感谢GPU(GPUs)。没有它们，使用计算机会变得不那么丰富多彩和友好！
HDG Explains: What Is a GPU?
Τhe GPU or Graphical Processing Unit is a dedicated microchip that is designed to specifically do graphics-related tasks at a much faster rate than a CPU or Central Processing Unit.
In general, GPUs handle the visual side of video games as well as video editing software, graphic design, 3D design and other similar tools. Recently, GPUs have also been put to use doing certain non-graphics tasks like cryptocurrency mining, which works well with their unique internal design.
The Difference Between a GPU And CPU
Every computer, whether a laptop, tablet or gaming console, has a CPU. The CPU is the brain of the machine. When you run software on your device, it’s the CPU that follows the instructions and does all the necessary math. Of course, every microchip executes instructions, so what makes a CPU special?
Most modern CPUs have four CPU cores, although this number is rapidly growing. The number of calculations a CPU can do at the same time is limited by the number of cores. But a CPU is very good at doing a small number of complex calculations of any type, very quickly, one after the other.
However, modern GPUs have a very different design philosophy. Instead of a small number of complex, full-function processing cores, they have hundreds or thousands of simple processor cores that are designed to execute instructions in parallel, specifically related to graphics.
For example, an instruction to shade a pixel in the image a certain way. Modern computer displays have millions of pixels. To render and animate realistic 3D graphics at smooth frame rates demands this massive parallel-processing power. So if you want to process lots and lots of simple commands at the same time, you want a GPU.
Are GPUs And Graphics Cards The Same Thing?
It’s become common to use the terms “GPU” and “graphics card” interchangeably. Which leads to situations where someone will say that their computer doesn’t have a GPU, but what they really mean is that the computer doesn’t have a graphics card.
What is a GPU? The GPU refers specifically to the microprocessor itself, but of course a GPU can’t operate by itself. It needs other components to do its job. So let’s go over the different ways that GPUs are packaged.
Graphics Cards
A graphics card is a discrete device that plugs into a standard slot on your computer motherboard. It consists of the printed circuit board, the GPU, memory, display connectors, power connectors and other various components that the GPU needs.
Graphics cards also have dedicated cooling systems that help keep the GPU within a safe operating temperature. The main benefit of graphics cards is that they can be easily replaced with newer models without the need to replace the entire computer or motherboard.
Integrated GPUs
Integrated GPUs are built into the same housing as a CPU. So a computer equipped with one of these does not need a separate graphics card in order to work. Display connectors are provided by the motherboard and the GPU shares the same pool of memory as the CPU. Of course, the CPU and GPU also share the same cooling system and power supply.
Integrated GPUs are pretty powerful these days and some can even be used for modest video game graphics, but their main advantage is cost and space savings. Which is why they are a popular solution in many laptops.
Tablets and smartphones also have integrated GPUs, but these are part of a “system on a chip” package. On computers, only the CPU and GPU share the same package In the case of SoCs, the RAM, storage, peripheral controllers and other components that make up the entire system are all on the same silicon package.
Dedicated GPUs
A dedicated GPU is distinct from a graphics card and integrated GPUs. It’s easiest to think of it as a graphics card that has been built into the computer. Laptops are the main devices that use dedicated GPUs.
Just as with a graphics card, the GPU has its own memory and cooling, but unlike a graphics card you can’t change or upgrade your dedicated GPU. There are a small number of laptops that use interchangeable GPU modules, but this is a rare exception.
External Graphics Cards
A recent development, external graphics cards, make use of high-speed ports on modern computers. Thunderbolt 3 over USB C is the solution of choice. External cards can either be an integrated solution, or simply an enclosure which you can use with any graphics card that will fit.
This is a good solution for laptop users who want a more powerful GPU solution when at a desk. You can read more about external GPUs in our article about them.
Understanding GPU Specifications
If you want to know how to choose between different graphics cards, have a look at our graphics card buyer’s guide for a detailed explanation of GPU specifications. If you just want a quick rundown, here are the key numbers that matter when it comes to GPUs:
- The number of processors
- The amount of memory
- The total memory bandwidth
- Special features, such as hardware ray tracing
Unfortunately, knowing these numbers or details won’t tell you much about how a given GPU or graphics card will perform. In the end what matters most is performance with the applications you want to run, which you can easily find in published benchmarks online.
General GPU Computing Tasks
The last important thing to know about modern GPUs is that they can be used for certain non-graphics tasks. Since GPUs are essentially made up of a massive number of simple processors, any job that needs lots of calculations done in parallel may benefit from running on a GPU.
For example, the Nvidia brand of cards work with CUDA (Compute Unified Device Architecture) instructions. You can write software using CUDA so that it makes use of the GPU’s unique design to accelerate certain types of calculations.
There’s also OpenCL, which is not attached to a specific brand of card. GPUs are being applied to big data analysis, machine learning, deep learning, cryptocurrency mining, and an ever-growing list of advanced jobs that would have needed supercomputers in the past.
I Can See Clearly Now
These are the most important facts you need to know about what GPUs are. Now you know what they are, what they’re used for, and how they’re different from other types of processors. We should all be grateful for GPUs. Using computers would be much less colorful and friendly without them!
