diff --git a/docs/src/gpu.md b/docs/src/gpu.md index fc97eed377..b52d9574c6 100644 --- a/docs/src/gpu.md +++ b/docs/src/gpu.md @@ -2,6 +2,17 @@ NVIDIA GPU support should work out of the box on systems with CUDA and CUDNN installed. For more details see the [CUDA](https://github.com/JuliaGPU/CUDA.jl) readme. +## Checking GPU Availability + +By default, Flux will run the checks on your system to see if it can support GPU functionality. You can check if Flux identified a valid GPU setup by typing the following: + +```julia +julia> using CUDA + +julia> CUDA.functional() +true +``` + ## GPU Usage Support for array operations on other hardware backends, like GPUs, is provided by external packages like [CUDA](https://github.com/JuliaGPU/CUDA.jl). Flux is agnostic to array types, so we simply need to move model weights and data to the GPU and Flux will handle it.