是否可以在gpu上完全执行应用程序（非图形化）？ [英] Is it possible to execute an application(not graphical) completely on gpu?

问题描述

所有计算机中都存在IS gpu，是否可以在GPU上完全执行应用程序（非图形化）？



我尝试过：



如果可能有任何代码样本可能在任何语言中

解决方案

Quote：

IS gpu存在于所有计算机中

很多是，全部否。

< blockquote class =quote>

引用：

是否可以在GPU上执行应用程序？

答案是否因为GPU是从属处理器，无法访问硬件。

Quote：

是否有可能在GPU上执行应用程序（非图形）？

GPU专为大规模并行计算而定制，因此高度并行的应用程序的任何部分都可以通过有效地卸载到GPU。

主CPU只是卸载计算机在GPU上，没有别的。简单地说GPU是通用的，它也处理非图形化的东西。



直到操作系统直接处理GPU使用情况，你总是会在主CPU中有东西来管理我/ O和用户交互。

不仅这是可能的，而且计算活动中存在称为GPGPU的整个字段：图形处理单元上的通用计算 - 维基百科，免费的百科全书 [ ^ ]。



当然，这种方法是计算领域技术发展的一个奇怪的悖论：最初为服务显示器设计的技术现在有时用于提供通用计算，提供比主机系统CPU更多的计算能力。但是，在实际操作中，这种方法用于许多领域。值得注意的是，利用GPU功能的应用程序仍允许它们在屏幕上显示图形，与GPGPU操作并行。



这是最常用的技术之一排序是NVIDIA CUDA：

CUDA - 维基百科，免费的百科全书 [ ^ ]，

并行编程和计算平台| CUDA | NVIDIA | NVIDIA [ ^ ]。



对于代码示例，几个快速答案的代码示例几乎无法帮助您。你必须很好地学习这个领域，以了解如何使用它。例如，对于CUDA，您可以找到大量材料，甚至可以学习包含代码示例和家庭作业的课程，例如： CS 179：GPU编程 [ ^ ]。



您需要独立研究该主题，达到合理的深度，以便能够提出更具体和合格的问题。

-SA

IS gpu present in all computer and is it possible to execute an application(Not graphical) completely on GPU?

What I have tried:

if it is possible any sample of code to do it that may be in any languague

解决方案

Quote:

IS gpu present in all computer

Many Yes, all No.

Quote:

is it possible to execute an application completely on GPU?

The answer is NO because the GPU is a slave processor and don't have access to hardware.

Quote:

is it possible to execute an application (Not graphical) on GPU?

The GPU is tailored for massively parallel calculus, so any part of an app that is heavily parallel can by efficiently off loaded to GPU.
The main CPU simply offload computing on the GPU, nothing else. Simply the GPU being general purpose, it also handle non graphical stuff.

Until the OS directly handle the GPU usage, you will always have something in main CPU to manage I/O and user interaction.

Not only this is possible, but there is whole field in computing activity called GPGPU: General-purpose computing on graphics processing units — Wikipedia, the free encyclopedia[^].

Of course, this approach is one of the weird paradoxes of technological development in the computing world: the technologies initially designed purely for serving displays are now sometimes used for serving up the general-purpose calculations, providing more computing power than the host system's CPUs. However, in real practice this approach is used in a number of areas. Notably, the application leveraging the power of GPUs, still allow them to show graphics on screen, in parallel to the GPGPU operation.

One of the most used technologies of this sort is NVIDIA CUDA:
CUDA — Wikipedia, the free encyclopedia[^],
Parallel Programming and Computing Platform | CUDA | NVIDIA|NVIDIA[^].

As to the code samples, few code sample from few Quick Answers hardly can help you. You have to learn this field well, to understand how to use it. For CUDA, for example, you can find a lot of material, and even study courses with code samples and homework assignments, like this one: CS 179: GPU Programming[^].

You need to study the subject independently, to a reasonable depth, to be able to ask more specific and qualified questions.

—SA

这篇关于是否可以在gpu上完全执行应用程序（非图形化）？的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！