sfcf slightly modified

src/LatticeGPU.jl

@@ -64,6 +64,6 @@ include("Solvers/Solvers.jl")
 using .Solvers
 export CG!
 export propagator, bndpropagator
-export fP!, fA!, kV!, lV!
+export fP, fA, kV, lV
 
 end # module

src/Solvers/Solvers.jl

@@ -26,7 +26,7 @@ include("Propagators.jl")
 export propagator, bndpropagator
 
 include("sfcf.jl")
-export fP!, fA!, kV!, lV!
+export fP, fA, kV, lV
 
 
 end

src/Solvers/sfcf.jl

 
 
 
-function fP!(fp, dws::DiracWorkspace, prop::AbstractArray,lp::SpaceParm)
+function fP(prop::AbstractArray,dws::DiracWorkspace, lp::SpaceParm)
 
     function krnl_fP!(sum,pro)
             b=Int64(CUDA.threadIdx().x)
@@ -16,12 +16,10 @@ function fP!(fp, dws::DiracWorkspace, prop::AbstractArray,lp::SpaceParm)
         CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_fP!(dws.cs,prop)
     end
 
-    fp += reshape(Array(CUDA.reduce(+, dws.cs; dims=(1,2,3))),lp.iL[end])
-
-    return nothing
+    return reshape(Array(CUDA.reduce(+, dws.cs; dims=(1,2,3))),lp.iL[end])
 end
 
-function fA!(fa, dws::DiracWorkspace, prop::AbstractArray,lp::SpaceParm)
+function fA(prop::AbstractArray,dws::DiracWorkspace, lp::SpaceParm)
 
     function krnl_fA!(sum,pro)
             b=Int64(CUDA.threadIdx().x)
@@ -35,11 +33,11 @@ function fA!(fa, dws::DiracWorkspace, prop::AbstractArray,lp::SpaceParm)
     CUDA.@sync begin
         CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_fA!(dws.cs,prop)
     end
-    fa += reshape(Array(CUDA.reduce(+, dws.cs; dims=(1,2,3))),lp.iL[end])
-    return nothing
+    
+    return reshape(Array(CUDA.reduce(+, dws.cs; dims=(1,2,3))),lp.iL[end])
 end
 
-function kV!(kv, dws::DiracWorkspace, prop::AbstractArray,lp::SpaceParm)
+function kV(prop::AbstractArray,dws::DiracWorkspace, lp::SpaceParm)
 
     function krnl_kV!(sum,pro)
             b=Int64(CUDA.threadIdx().x)
@@ -53,11 +51,11 @@ function kV!(kv, dws::DiracWorkspace, prop::AbstractArray,lp::SpaceParm)
     CUDA.@sync begin
         CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_kV!(dws.cs,prop)
     end
-    kv += reshape(Array(CUDA.reduce(+, dws.cs; dims=(1,2,3))),lp.iL[end])
-    return nothing
+    
+    return reshape(Array(CUDA.reduce(+, dws.cs; dims=(1,2,3))),lp.iL[end])
 end
 
-function lV!(lv, dws::DiracWorkspace, prop::AbstractArray,lp::SpaceParm)
+function lV(prop::AbstractArray,dws::DiracWorkspace, lp::SpaceParm)
 
     function krnl_lV!(sum,pro)
             b=Int64(CUDA.threadIdx().x)
@@ -73,6 +71,6 @@ function lV!(lv, dws::DiracWorkspace, prop::AbstractArray,lp::SpaceParm)
     CUDA.@sync begin
         CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_lV!(dws.cs,prop)
     end
-    lv += reshape(Array(CUDA.reduce(+, dws.cs; dims=(1,2,3))),lp.iL[end])
-    return nothing
+    
+    return reshape(Array(CUDA.reduce(+, dws.cs; dims=(1,2,3))),lp.iL[end])
 end