# BFloat16 on ARM for neural networks

**URL:** <https://discourse.julialang.org/t/bfloat16-on-arm-for-neural-networks/28200>\
**Category:** Machine Learning\
**Created:** [August 30, 2019, 8:44am UTC](https://discourse.julialang.org/t/bfloat16-on-arm-for-neural-networks/28200 "2019-08-30T08:44:28Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![johnh](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/johnh/32/3615_2.png) [@johnh](https://discourse.julialang.org/u/johnh)\
**Post date:** [August 30, 2019, 8:44am UTC](https://discourse.julialang.org/t/bfloat16-on-arm-for-neural-networks/28200/1 "2019-08-30T08:44:29Z")

</div>

> **[BFloat16 extensions for Armv8-A](https://community.arm.com/arm-community-blogs/b/ai-and-ml-blog/posts/bfloat16-processing-for-neural-networks-on-armv8_2d00_a)**
>
> The next revision of the Armv8-A architecture will introduce Neon and SVE vector instruction designed to accelerate Neural Networks using the BFloat16 format.

How does this play with Julia? Someone will now tell me thereis a BFlaot16 type in core already!

---

<div class="post-metadata">

**Author:** ![ChrisRackauckas](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/chrisrackauckas/32/77_2.png) [@ChrisRackauckas](https://discourse.julialang.org/u/ChrisRackauckas)\
**Post date:** [August 30, 2019, 9:56am UTC](https://discourse.julialang.org/t/bfloat16-on-arm-for-neural-networks/28200/2 "2019-08-30T09:56:19Z")

</div>

[https://github.com/JuliaComputing/BFloat16s.jl](https://github.com/JuliaComputing/BFloat16s.jl)

Was created for testing TPU code.