Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

切分的颗粒度是多少? #470

Open
janetat opened this issue Sep 5, 2024 · 4 comments
Open

切分的颗粒度是多少? #470

janetat opened this issue Sep 5, 2024 · 4 comments

Comments

@janetat
Copy link

janetat commented Sep 5, 2024

问题

  1. 算力、显存的切割颗粒度是多少呢?是否支持1%算力切割。
  2. 怎么验证算力切割颗粒度
@archlitchi
Copy link
Collaborator

device memory: 1M
core util:1%

you can run a tensorflow-benchmark and compare the images/s result between 100% util and 50% util

@janetat
Copy link
Author

janetat commented Sep 5, 2024

  1. 如果设置的是1%的算力限制的话,tensorflow-benchmark能测试出来么?
  2. 是一个周期内平均下来的吞吐率是1%吗?

@archlitchi
Copy link
Collaborator

  1. 如果设置的是1%的算力限制的话,tensorflow-benchmark能测试出来么?
  2. 是一个周期内平均下来的吞吐率是1%吗?

yes to 2 questions

@jessehu
Copy link

jessehu commented Oct 7, 2024

hi @archlitchi, 请问是否存在这样的算力控制现象:GPU算力单元的利用率会超过设置的值(比如单卡切分为2卡,显存是控制住了50%,但某一张虚拟卡的算力利用率会在一些小时间段内超过50%)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants