JuliaGPU / ArrayFire.jl

Julia wrapper for the ArrayFire library
Other
204 stars 37 forks source link

ArrayFire "Failed to load dynamic library" on Windows #267

Open denkab opened 3 years ago

denkab commented 3 years ago

HI!

I'm trying to enable ArrayFire with Julia on older machine with CPU backend, so I'm making it "simple" for ArrayFire to choose by setting ENV JULIA_ARRAYFIRE_BACKEND=CPU.

It's not working still...

julia> using ArrayFire ERROR: InitError: ArrayFire Error (501) : Failed to load dynamic library. Error in af_err __cdecl unified::setBackend(af_backend) In file src\api\unified\symbol_manager.cpp:245: Failed to load dynamic library. See http://www.arrayfire.com/docs/unifiedbackend.htm for instructions to set up environment for Unified backend. 0# af_is_real in af 1# af_is_real in af 2# 0x000000002D59CF3E

What can I try to solve this?

If I don't set the backend to CPU, the error is

ERROR: InitError: ArrayFire Error (501) : Failed to load dynamic library. Error in af_err __cdecl af_init(void) In file src\api\unified\device.cpp:47: ArrayFire couldn't locate any backends. 0# af_is_real in af 1# af_is_real in af 2# 0x000000002D60CF71

denkab commented 3 years ago

Possibly worth mentioning that OpenCL.jl is functional on the same set up, reporting CPU as the only available backend,

julia> cl.platforms() 1-element Array{OpenCL.cl.Platform,1}: OpenCL.Platform('AMD Accelerated Parallel Processing' @0x00007ffe95ebe188)