# DL4J and Mobile GPU

**URL:** <https://community.konduit.ai/t/dl4j-and-mobile-gpu/912>\
**Category:** Feature Request\
**Created:** [October 7, 2020, 1:20pm UTC](https://community.konduit.ai/t/dl4j-and-mobile-gpu/912 "2020-10-07T13:20:18Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![lavajaw](https://yyz1.discourse-cdn.com/flex035/user_avatar/community.konduit.ai/lavajaw/32/118_2.png) [@lavajaw](https://community.konduit.ai/u/lavajaw)\
**Post date:** [October 7, 2020, 1:20pm UTC](https://community.konduit.ai/t/dl4j-and-mobile-gpu/912/1 "2020-10-07T13:20:19Z")

</div>

Hello,

Currnetly I am developing DL4J models for Android applications. My models must be very light because all computations are done on CPU. Do you have any plans to make somekind mobile GPU support? I think that this option is very important.

Thanks on your time,  
Aleksandar

---

<div class="post-metadata">

**Author:** ![agibsonccc](https://yyz1.discourse-cdn.com/flex035/user_avatar/community.konduit.ai/agibsonccc/32/697_2.png) [@agibsonccc](https://community.konduit.ai/u/agibsonccc)\
**Post date:** [October 9, 2020, 4:51am UTC](https://community.konduit.ai/t/dl4j-and-mobile-gpu/912/2 "2020-10-09T04:51:26Z")

</div>

@lavajaw yeah we’re working on that. We need to finish up our ARM additions:

> <https://github.com/KonduitAI/deeplearning4j/pull/503>
>
> \## What changes were proposed in this pull request?
> Ops Additions:
> \- added con…v2d (float32)
> \- added deconv2d (float32)
> \- minor fixes of coding-style and so on
> 
> libnd4j cross-compilation script for pi(arm):  
> 
> \- use new arm cross compiler from the official arm site
> \- minor fixes in CmakeLists
> \- fix DEBUG compilation by manually switching linker to // ld.gold --long-plt //, because of faulty gcc -fuse-ld=gold switch
> 
> ShapeDescriptor:
> \- validation
> \- paddedBuffer helper
> \- proper allocLength ~~and other helper member methods~~
> \- the flag indicating that Array having padded buffer
> 
> NdArray Factory methods
> \- generic flexible interface for creating NdArray
> \- generic flexible interface for padded NdArrays. As it is more generic it's easier to simulate arm\_compute {top, right, bottom, left} padding with it.
> 
> \## How was this patch tested?
> 
> 
> unit tests
> 
> \## Quick checklist
> 
> The following checklist helps ensure your PR is complete:
> 
> \- \[x\] Eclipse Contributor Agreement signed, and signed commits - see \[IP Requirements\](https://deeplearning4j.org/eclipse-contributors) page for details
> \- \[x\] Reviewed the \[Contributing Guidelines\](https://github.com/eclipse/deeplearning4j/blob/master/CONTRIBUTING.md) and followed the steps within.
> \- \[x\] Created tests for any significant new code additions.
> \- \[x\] Relevant tests for your changes are passing.

We put that on hold a bit but will be picking it back up soonish.