add MPI.jl dep

Author: mofeing

Project.toml

@@ -31,6 +31,7 @@ CUDA = "052768ef-5323-5732-b1bb-66c8b64840ba"
 GPUCompiler = "61eb1bfa-7361-4325-ad38-22787b887f55"
 KernelAbstractions = "63c18a36-062a-441e-b654-da1e3ab1ce7c"
 LLVM = "929cbde3-209d-540e-8aea-75f648917ca0"
+MPI = "da04e1cc-30fd-572f-bb4f-1f8673147195"
 NNlib = "872c559c-99b0-510c-b3b7-b6c96a88d5cd"
 OffsetArrays = "6fe1bfb0-de20-5000-8ca7-80f57d26f881"
 PythonCall = "6099a3de-0909-46bc-b1f4-468b9a2dfc0d"