can i use pthreads to program Nvidia GPUs?

Accelerated Computing CUDA CUDA Programming and Performance
#1

Hi,
I am planning to buy a machine having about 1300 cores on Nvidia GPU.

I have experience programming with pthreads.

If I create a new thread using pthread_create and give it a function to execute, will it
automatically use one of my 1300 cores on GPU to execute it?

It is as simple as above for multiple cores existing on a CPU.
Will it be more complex for GPU programming?

Thanks much,
mani

#2

No. To know why, I really recommend reading any, any of the abundant documentation available.

#3

我认为是不可以的,pthread是Linux系统的多线程编程接口,cuda是GPU多线程,底层硬件就不一样.不光是pthread不能使用,支持CPU的绝大部分函数都不可以使用,比如std::cout 就不行.个人理解是CPU的代码是不可以用在GPU上的.除非NVIDIA公司重载了CPU上的函数,GPU本质上就是一个做运算的硬件.支持C++ 语言,个人理解是只是语法上支持,单并不支持现有的C++函数.