# Example / Image for EC2 GPU with Julia?

**URL:** <https://discourse.julialang.org/t/example-image-for-ec2-gpu-with-julia/21124>\
**Category:** General Usage\
**Created:** [February 23, 2019, 9:46pm UTC](https://discourse.julialang.org/t/example-image-for-ec2-gpu-with-julia/21124 "2019-02-23T21:46:00Z")\
**Posts on this page:** 5\
**Page:** 1

<div class="post-metadata">

**Author:** ![philip](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/philip/32/17191_2.png) [@philip](https://discourse.julialang.org/u/philip)\
**Post date:** [February 23, 2019, 9:46pm UTC](https://discourse.julialang.org/t/example-image-for-ec2-gpu-with-julia/21124/1 "2019-02-23T21:46:00Z")

</div>

I’m wondering if anyone with experience with GPU programming has examples of spinning up Amazon EC2 instances with GPUs and successfully running Julia on them. I take it the package installation process for getting Julia to “talk” to a GPU can be a bit of a bear. I’m hoping to do something like a Value Function Iteration akin to that demonstrated in this JuliaCon talk:

[![](https://global.discourse-cdn.com/julialang/original/3X/2/5/25c04238d804df182ad6cd0a6796f727b3d484b9.jpeg "GPU - Accelerated Value Function Iteration in Julia | Giovanni Ballarin") ](https://www.youtube.com/watch?v=ehUb-S6Bxnk)

I see excellent example code here that looks as though it assumes an already configured machine, but I’m wondering if there is any accumulated knowledge from which I could benefit around the stage of configuring the machine. In the past I’ve found getting EC2 nodes working to be a grueling process.

> **[GitHub - giob1994/GPU-OpenCL-VFI-in-Julia: GPU + OpenCL Value Function...](https://github.com/giob1994/GPU-OpenCL-VFI-in-Julia)**
>
> GPU + OpenCL Value Function Iteration for Macro Models - GitHub - giob1994/GPU-OpenCL-VFI-in-Julia: GPU + OpenCL Value Function Iteration for Macro Models

---

<div class="post-metadata">

**Author:** ![xiaodai](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/xiaodai/32/15937_2.png) [@xiaodai](https://discourse.julialang.org/u/xiaodai)\
**Post date:** [February 24, 2019, 2:39am UTC](https://discourse.julialang.org/t/example-image-for-ec2-gpu-with-julia/21124/2 "2019-02-24T02:39:05Z")

</div>

CuArrays are easier. It’s based on CUDA not CLArrays

---

<div class="post-metadata">

**Author:** ![Tomas\_Pevny](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/tomas_pevny/32/25466_2.png) [@Tomas\_Pevny](https://discourse.julialang.org/u/Tomas_Pevny)\
**Post date:** [February 24, 2019, 11:09am UTC](https://discourse.julialang.org/t/example-image-for-ec2-gpu-with-julia/21124/3 "2019-02-24T11:09:44Z")

</div>

I do not have an experience, but I have just used Amazons ami for GPU and deep learning, install Flux, CuArrays and everything worked out of the box without any hassle.

---

<div class="post-metadata">

**Author:** ![philip](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/philip/32/17191_2.png) [@philip](https://discourse.julialang.org/u/philip)\
**Post date:** [February 25, 2019, 1:31pm UTC](https://discourse.julialang.org/t/example-image-for-ec2-gpu-with-julia/21124/4 "2019-02-25T13:31:57Z")

</div>

Cool; thanks. I’ll give it a shot.

---

<div class="post-metadata">

**Author:** ![pkofod](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/pkofod/32/2179_2.png) [@pkofod](https://discourse.julialang.org/u/pkofod)\
**Post date:** [February 25, 2019, 2:00pm UTC](https://discourse.julialang.org/t/example-image-for-ec2-gpu-with-julia/21124/5 "2019-02-25T14:00:22Z")

</div>

> [@Tomas\_Pevny](#):
>
> I do not have an experience, but I have just used Amazons ami for GPU and deep learning, install Flux, CuArrays and everything worked out of the box without any hassle.

What version of Julia? 1.1? I was under the impression that CuArrays did not work on v1.1