# About using partial GPU

**URL:** <https://discuss.ray.io/t/about-using-partial-gpu/113>\
**Category:** Ray Core\
**Created:** [December 2, 2020, 2:47pm UTC](https://discuss.ray.io/t/about-using-partial-gpu/113 "2020-12-02T14:47:37Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![kyunghyun.lee](https://sea2.discourse-cdn.com/flex020/user_avatar/discuss.ray.io/kyunghyun.lee/32/25_2.png) [@kyunghyun.lee](https://discuss.ray.io/u/kyunghyun.lee)\
**Post date:** [December 2, 2020, 2:47pm UTC](https://discuss.ray.io/t/about-using-partial-gpu/113/1 "2020-12-02T14:47:37Z")

</div>

Hi,

I want to use “partial GPUs” (not fractional GPU)  
For example, I have 8 GPU in one machine.  
I want to assign one job (including many actors and sub-jobs) to GPU 0-3 and assign another job (with hyperparameter changed) to GPU 4-8, similar to other deep learning frameworks using “CUDA\_VISIBLE DEVICES.”  
How can I set ray to achieve this?

Thanks!

---

<div class="post-metadata">

**Author:** ![rliaw](https://sea2.discourse-cdn.com/flex020/user_avatar/discuss.ray.io/rliaw/32/24_2.png) [@rliaw](https://discuss.ray.io/u/rliaw)\
**Post date:** [December 4, 2020, 8:17pm UTC](https://discuss.ray.io/t/about-using-partial-gpu/113/2 "2020-12-04T20:17:56Z")

</div>

Hmm, interesting!

Will this work if you just do `ray.remote(num_gpus=4)` on each function?