nvidia-smi内存使用与GPU内存使用之间有何区别? [英] What's the difference between nvidia-smi Memory-Usage and GPU Memory Usage?

问题描述

例如，运行nvidia-smi命令将显示:

Running the nvidia-smi command displays for example:

Memmory-Usage和GPU Memory Usage有什么区别?

推荐答案

Memory-Usage表示:

used memory/total memory

已用内存是指GPU正在使用的内存量.如果没有人使用总内存，则总内存就是可用内存.

Used memory is how much of the GPU's memory is in use. Total Memory is how much is available, if nobody were using any of it.

GPU Memory Usage只是按流程将其分解.

GPU Memory Usage simply breaks this down by process.

该输出显然有一个进程正在运行，并且该进程正在使用10557 MB的GPU 0内存.这与上面报告的10570数字基本相同.是的，差异很小，这是由于GPU内存消耗与该进程无关(例如CUDA本身正在使用).

This output apparently has one process running, and that process is using 10557 megabytes of GPU 0 memory. That is essentially the same as the 10570 number reported above it. Yes, there is a small difference, it is due to GPU memory consumption not associated with that process (e.g. in use by CUDA itself).

这篇关于nvidia-smi内存使用与GPU内存使用之间有何区别?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！