Cuda literacy sticker

Q: What is Cuda?
A: Cuda is a NVIDIA gpgpu model. Based on the C language, it can be directly written into the C language that most people are familiar with and executed on the display chip.ProgramInstead of learning specific display chip commands or special structures.

Q: Why does NVIDIA develop Cuda?
A: You have heard of gpgpu before. It is a general-purpose GPU. With the complexity of 3D graphics processing, the GPU scale is getting bigger and bigger. Later, from the computing point of view, performance is much faster than CPU performance. GPU computing capabilities are so powerful that in addition to graphic processing, it can also be used for other applications, that is, general-purpose computing.
At that time, it was very troublesome to use GPU computing. At that time, GPU was completely designed for graphics. It was a very dedicated processor. If you want to use GPU for computing, you need to use the graphic API to access the GPU computing core. You must use the API to assume that the computing is 3D, then you have to go through the graphic processing process to get the calculation result. First, programming is very complex, and second, efficiency is not very high. Because the rendering process must be passed in the middle.
However, for those who really need parallel computing, such as scientists, it may be difficult for them to learn graphics again. These people are all busy. In this case, parallel computing and graphics are combined, so NVIDIA developed Cuda.


Q: What is the value of Cuda for the industry?
A: Cuda supports cudagpu for the value of the industry. When the two are combined, it is easy for all vendors to use the powerful computing capabilities of GPU to perform a variety of parallel computing tasks, this is the maximum price. It does not require you to understand anything else. You only need to have the C Programming capability. C language is the most popular language.


Q: What kind of groups can Cuda address?
A: Cuda is not only intended for software developers, but also for various programs. For example, people playing games can also enjoy many benefits, that is, computing on physical acceleration.


Q: Can I benefit from the Cuda system if I do not use the Nv product? For example, if my current computer does not use NV hardware, can I develop it based on the Cuda platform?
A: There is no NVIDIA graphics card. You can also use Cuda for development. You can debug the Development and use a virtual graphics card. In addition, Cuda supports multi-core CPUs. This is our latest support. You can write on the CPU. If you have a GPU, it is much faster than the CPU, and our latest support is already on a multi-core CPU. Without a GPU, the extended program can also be compiled. One of the biggest advantages of Cuda is that, unlike other parallel computing, Cuda hardware costs are very low. Currently, Cuda costs 200 or 300 RMB a lot in the market, and the speed is about 80 million, it can be performed on many PCs, either workstation or server.


Q: What is the difference between Cuda and other C "> compiling environments">? What is the architecture of Cuda programming? What is the difference from other graphic programming platforms?
A: Cuda is a platform dedicated to GPU programming. Its biggest feature is that it is a so-called Heterogeneous Computing System in terms of GPU. It is very different from the CPU. the CPU is only for one processor, and Cuda is for the GPU. During editing Code Separate from GPU code, GPU code is compiled into work code, and the CPU still needs to be compiled by other C language compiling systems. This may be the biggest difference. Cuda must involve the CPU, which is called heterogeneous computing. The so-called heterogeneous computing is that the CPU and GPU architectures are different, but they work together to solve the same problem.
As for the relationship between Cuda and DirectX, OpenGL is not required for Cuda.


Q: Which operating systems are currently supported for graphics cards that contain the Cuda architecture? That is to say, can I compile parallel computing programs running on Linux or Unix systems?
A: Windows Vista and XP operating systems of the main stream, the latest win7, Linux and UNIX operating systems, and almost all operating systems are included. Cuda2.0beta has been embedded into the driver of the video card. You can use a variety of Cuda programs when using Cuda.


Q: Is the Cuda library compatible with the C language library such as ATL? Does Cuda have a fault tolerance and validation mechanism?
A: The Cuda library is different from the standard template library of C language, and Cuda will have a fault tolerance and validation mechanism in the future.