search

AMReX-Fluids / IAMR

A parallel, adaptive mesh refinement (AMR) code that solves the variable-density incompressible Navier-Stokes equations.
https://amrex-fluids.github.io/IAMR/
80 stars 57 forks source link

Use GpuArray instead of C array #78

Closed WeiqunZhang closed 3 years ago

WeiqunZhang commented 3 years ago

The array type can be lambda captured to device by recent versions of CUDA. However, it does not work with HIP and DPC++. So we switch to GpuArray instead.

cgilet commented 3 years ago

@WeiqunZhang @drummerdoc We want the target of this to be development, right?