search

omlins / ParallelStencil.jl

Package for writing high-level code for parallel high-performance stencil computations that can be deployed on both GPUs and CPUs
BSD 3-Clause "New" or "Revised" License
311 stars 31 forks source link

Caching the GPU kernels? #85

Closed korbinian90 closed 1 year ago

korbinian90 commented 1 year ago

After starting Julia, the first usage / compilation of the GPU kernels takes about 30s for me. This is done everytime when I restart Julia. Is there a possibility that this is cached? Since Julia 1.9, it seems to me like most compilation is cached, it would be great, if this is possible for this package as well.

omlins commented 1 year ago

@korbinian90 thanks for reporting this issue. However, for caching of compiled code we rely on the packages we use for the backends as CUDA.jl and AMDGPU.jl, and Julia itself of course. Therefore, please open issues regarding compilation caching directly in these packages.