diffeqgpu_setup {diffeqr}

Setup DiffEqGPU

Description

This function initializes the DiffEqGPU package for GPU-parallelized ensembles. The first time will be long since it includes precompilation.

Usage

diffeqgpu_setup(backend)

Arguments

backend

the backend for the GPU computation. Choices are "CUDA", "AMDGPU", "Metal", or "oneAPI"

Value

Returns a degpu object which holds the module state of the Julia-side DiffEqGPU package. The core use is to use degpu$EnsembleGPUKernel() for choosing the GPU dispatch in the solve.

Examples

## Not run:  ## diffeqgpu_setup() is time-consuming and requires Julia+DifferentialEquations.jl

degpu <- diffeqr::diffeqgpu_setup(backend)


## End(Not run)