如何更改CUDA核心的数量以计算效率和可扩展性？ [英] How to change the number of CUDA cores for calculating the efficiency and scalability?

问题描述

我是CUDA的新手，我想问你的帮助，知道是否可以改变内核的数量来计算程序的效率和可扩展性，除了修改内核与每个块的线程数和用于执行的线程块。也就是说，如果假设效率和可扩展性基于处理器数量，而且我在FERMI GPU中开始使用512个核心，那么我如何从512更改为256,128等核心，以获得不同的加速比给定数据加载？
提前感谢！

I'm new on CUDA and I would like to ask your help to know if it's possible to change the number of cores to calculate the efficiency and scalability of a program, besides modifying the kernel with the number of threads per block and blocks of threads for execution. That is, if it's assumed that efficiency and scalability are based on the number of processors and I start with 512 cores in my FERMI GPU, How could I change from 512 to 256, 128, etc. cores in order to get differents Speedup's for a given data load?? Thanks in advance!!

推荐答案

CUDA不支持这种类型的硬件粒度，控制内核的方式，你问的问题。

CUDA doesn't support that type of hardware granularity and there is no way to control cores in the way you are asking about.

这篇关于如何更改CUDA核心的数量以计算效率和可扩展性？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！