Tracking the issues we have getting OpenCL support.
- [x]
adaptintegration: https://github.com/JuliaGPU/CLArrays.jl/pull/16 - [x] Dual number support: https://github.com/JuliaGPU/CLArrays.jl/issues/17
- [ ] Integration with
Flux.gpu - [ ] GPU-compatible softmax: https://github.com/JuliaGPU/GPUArrays.jl/issues/96
- [ ] Convolutions: https://github.com/JuliaGPU/GPUArrays.jl/issues/102
Worth checking out ROCm and MIOpen as part of this.
MikeInnes
All 9 comments
@simondanisch
ViralBShah
on 19 Feb 2018
on it :) most is done, just waiting for tags & tests to turn green
SimonDanisch
on 19 Feb 2018
SimonDanisch
on 19 Feb 2018
Hello! Does anyone know what the status of OpenCL support is?
yblainm
on 28 Apr 2019
CLArrays is broken due to Transpiler not being updated for Julia 1.0 (which is unlikely to happen unless some adventurous individual cares to take that on), so automatic conversion of Julia code (i.e. Flux models and functions) is not currently possible.
jpsamaroo
on 28 Apr 2019
Is ArrayFire a potential alternative that could be used as an underlying representation for OpenCL support?
dented42
on 21 May 2019
Yes, supporting ArrayFire arrays may be a good idea as well. I'm not sure what they're like convolutions wise but the basics should be straightforward.
MikeInnes
on 4 Jun 2019
FYI for anyone still watching this thread, #938 will address this for AMD GPUs by skipping OpenCL and using ROCArrays instead.
jpsamaroo
on 19 Nov 2019
I'll close this issue since it was pretty tied to CLArrays, and that work isn't moving any more. We can track ROCm support in #938.
MikeInnes
on 20 Nov 2019
Related issues
ExpandingMan
路
6Comments
aminya
路
4Comments
jw3126
路
6Comments
ssfrr
路
5Comments
pylat
路
3Comments
Most helpful comment
CLArrays is broken due to Transpiler not being updated for Julia 1.0 (which is unlikely to happen unless some adventurous individual cares to take that on), so automatic conversion of Julia code (i.e. Flux models and functions) is not currently possible.