use static thread scheduling (#260)

* use static thread scheduling

* Test Multithreading

* test with single and multiple threads

---------

Co-authored-by: Fabian Gans <fgans@bgc-jena.mpg.de>

.github/workflows/CI.yml

@@ -39,6 +39,8 @@ jobs:
             ${{ runner.os }}-
       - uses: julia-actions/julia-buildpkg@v1
       - uses: julia-actions/julia-runtest@v1
+        env:
+          JULIA_NUM_THREADS: '4'
       - uses: julia-actions/julia-processcoverage@v1
       - uses: codecov/codecov-action@v1
         with:

src/DAT/DAT.jl

@@ -1159,7 +1159,7 @@ using Base.Cartesian
             )
         end
     else
-        Threads.@threads for cI in CartesianIndices(map(i -> 1:length(i), loopRanges))
+        Threads.@threads :static for cI in CartesianIndices(map(i -> 1:length(i), loopRanges))
             innercode(
                 f,
                 cI,

test/DAT/DAT.jl

@@ -48,4 +48,18 @@
         xout .= sum(skipmissing(xin))
     end
     @test any(ismissing.(nofilter.data)) == false
+end
+
+@testset "Threaded input caches" begin
+    a = rand(1000,5)
+    b = YAXArray(a)
+    mapslices(b,dims="Dim_2") do x
+        xc = copy(x)
+        for i in 1:100
+            xc[1] = xc[1]
+            yield()
+            x == xc || error("Multithreaded buffer is not correct")
+        end
+        1.0
+    end
 end

test/DAT/mapcube.jl

@@ -15,6 +15,10 @@
             xout .= x1 .+ x2
         end
         @test r.data == a2.data .+ reshape(a3.data,(4,1,5))
+        r = mapCube((a2, a3), indims=(indims, indims), outdims=outdims,nthreads = [1]) do xout, x1, x2
+            xout .= x1 .+ x2
+        end
+        @test r.data == a2.data .+ reshape(a3.data,(4,1,5))
     end
 
     @testset "Reduction to empty outputs" begin