# Runtime ND4J Backend type selection

**URL:** <https://community.konduit.ai/t/runtime-nd4j-backend-type-selection/902>\
**Category:** ND4J\
**Created:** [October 2, 2020, 2:05pm UTC](https://community.konduit.ai/t/runtime-nd4j-backend-type-selection/902 "2020-10-02T14:05:11Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![bakcsa83](https://avatars.discourse-cdn.com/v4/letter/b/c37758/32.png) [@bakcsa83](https://community.konduit.ai/u/bakcsa83)\
**Post date:** [October 2, 2020, 2:05pm UTC](https://community.konduit.ai/t/runtime-nd4j-backend-type-selection/902/1 "2020-10-02T14:05:11Z")

</div>

Hi,

I have a use case where I would like to use both CPU and CUDA backends. Backend type for a specific job should be decided based on job type, job size, available memory for the backends and a couple of other factors.  
Is it possible to tell ND4J at runtime which backend to be used for the next INDArray declaration?