# Specify in rasa to access CPU

**URL:** <https://forum.rasa.com/t/specify-in-rasa-to-access-cpu/22630>\
**Category:** Rasa Open Source\
**Created:** [December 19, 2019, 9:17am UTC](https://forum.rasa.com/t/specify-in-rasa-to-access-cpu/22630 "2019-12-19T09:17:56Z")\
**Posts on this page:** 1\
**Showing post:** 2

<div class="post-metadata">

**Author:** ![Arjaan](https://dub1.discourse-cdn.com/flex013/user_avatar/forum.rasa.com/arjaan/32/5754_2.png) [@Arjaan](https://forum.rasa.com/u/Arjaan)\
**Post date:** [January 15, 2020, 1:52am UTC](https://forum.rasa.com/t/specify-in-rasa-to-access-cpu/22630/2 "2020-01-15T01:52:30Z")

</div>

Hi @akanksha,

Rasa is not specifying what GPUs to use when it it uses Tensorflow. It will always use whatever GPUs are available.

If you want to control what GPUs are being used, the best practice is to set the environment variable `CUDA_VISIBLE_DEVICES`.

These 2 references are very good to get more details:

- [CUDA Pro Tip: Control GPU Visibility with CUDA\_VISIBLE\_DEVICES](https://devblogs.nvidia.com/cuda-pro-tip-control-gpu-visibility-cuda_visible_devices/)
- [Stack Overflow solution](https://stackoverflow.com/a/39661999/5480536)

---

_[View the full topic](https://forum.rasa.com/t/specify-in-rasa-to-access-cpu/22630)._
