Added JACCASYNC API for threads.jl backend. Note that JACC.Async work…

…s as regular JACC on threads.jl backend. The idea of JACC.Async is to enable JACC for multi-device (GPUs) concurrent executions.

src/JACC.jl

@@ -20,6 +20,9 @@ using .BLAS
 include("JACCMULTI.jl")
 using .Multi
 
+include("JACCASYNC.jl")
+using .Async
+
 include("JACCEXPERIMENTAL.jl")
 using .Experimental
 

src/JACCASYNC.jl

@@ -0,0 +1,61 @@
+module Async
+
+using JACC
+import JACC: ThreadsBackend
+
+function Array(::ThreadsBackend, queue_id::I, x::Base.Array{T,N}) where {I <: Integer, T, N}
+    return JACC.Array(x)
+end
+
+function copy(::ThreadsBackend, queue_id_dest::I, x::Base.Array{T,N}, queue_id_orig::I, y::Base.Array{T,N}) where {I <: Integer, T, N}
+    copyto!(x, y)
+end
+
+function parallel_for(::ThreadsBackend, queue_id::I, N::I, f::F, x...) where {I <: Integer, F <: Function}
+    JACC.parallel_for(N, f, x...)
+end
+
+function parallel_for(::ThreadsBackend, queue_id::I, (M, N)::Tuple{I,I}, f::F, x...) where {I <: Integer, F <: Function}
+    JACC.parallel_for((M, N), f, x...)
+end
+
+function parallel_reduce(::ThreadsBackend, queue_id::I, N::I, f::F, x...) where {I <: Integer, F <: Function}
+    return JACC.parallel_reduce(N, f, x...)
+end
+
+function parallel_reduce(::ThreadsBackend, queue_id::I, (M, N)::Tuple{I, I}, f::F, x...) where {I <: Integer, F <: Function}
+    return JACC.parallel_reduce((M, N), f, x...)
+end
+
+function synchronize(::ThreadsBackend)
+end
+
+function Array(queue_id::I, x::Base.Array{T,N}) where {I <: Integer, T, N}
+    return Array(JACC.default_backend(), queue_id, x)
+end
+
+function copy(queue_id_dest::I, x::Base.Array{T,N}, queue_id_orig::I, y::Base.Array{T,N}) where {I <: Integer, T, N}
+    return copy(JACC.default_backend(), queue_id_dest, x, queue_id_orig, y)
+end
+
+function parallel_for(queue_id::I, N::I, f::F, x...) where {I <: Integer, F <: Function}
+    return parallel_for(JACC.default_backend(), queue_id, N, f, x...)
+end
+
+function parallel_for(queue_id::I, (M, N)::Tuple{I, I}, f::F, x...) where {I <: Integer, F <: Function}
+    return parallel_for(JACC.default_backend(), queue_id, (M, N), f, x...)
+end
+
+function parallel_reduce(queue_id::I, N::I, f::F, x...) where {I <: Integer, F <: Function}
+    return parallel_reduce(JACC.default_backend(), queue_id, N, f, x...)
+end
+
+function parallel_reduce(queue_id::I, (M, N)::Tuple{I, I}, f::F, x...) where {I <: Integer, F <: Function}
+    return parallel_reduce(JACC.default_backend(), queue_id, (M, N), f, x...)
+end
+
+function synchronize()
+    return synchronize(JACC.default_backend())
+end
+
+end # module Async