Add sorting, sparse constructor, and device ownership

[michel2323]

• Adding sort! and sortperm
• Add explicit device(x) ownership in global queue calls of MKL
• Add oneSparseMatrixCSR(rowptr, colptr, nzval, dims) constructor

[github-actions]

Your PR requires formatting changes to meet the project's style guidelines.
Please consider running Runic (git runic ms/sparse_fix) to apply these changes.

Click here to view the suggested changes.

diff --git a/lib/mkl/wrappers_sparse.jl b/lib/mkl/wrappers_sparse.jl
index 3f8b0bb..bb17907 100644
--- a/lib/mkl/wrappers_sparse.jl
+++ b/lib/mkl/wrappers_sparse.jl
@@ -1,5 +1,5 @@
 function sparse_release_matrix_handle(A::oneAbstractSparseMatrix)
-    if A.handle !== nothing
+    return if A.handle !== nothing
         try
             queue = global_queue(context(A.nzVal), device(A.nzVal))
             handle_ptr = Ref{matrix_handle_t}(A.handle)
@@ -8,7 +8,7 @@ function sparse_release_matrix_handle(A::oneAbstractSparseMatrix)
             synchronize(queue)
         catch err
             # Don't let finalizer errors crash the program
-            @warn "Error releasing sparse matrix handle" exception=err
+            @warn "Error releasing sparse matrix handle" exception = err
         end
     end
 end
@@ -117,10 +117,10 @@ for (fname, elty, intty) in ((:onemklSsparse_set_coo_data   , :Float32   , :Int3
             queue = global_queue(context(nzVal), device(nzVal))
             if m != 0 && n != 0
                 $fname(sycl_queue(queue), handle_ptr[], m, n, nnzA, 'O', rowInd, colInd, nzVal)
-                dA = oneSparseMatrixCOO{$elty, $intty}(handle_ptr[], rowInd, colInd, nzVal, (m,n), nnzA)
+                dA = oneSparseMatrixCOO{$elty, $intty}(handle_ptr[], rowInd, colInd, nzVal, (m, n), nnzA)
                 finalizer(sparse_release_matrix_handle, dA)
             else
-                dA = oneSparseMatrixCOO{$elty, $intty}(nothing, rowInd, colInd, nzVal, (m,n), nnzA)
+                dA = oneSparseMatrixCOO{$elty, $intty}(nothing, rowInd, colInd, nzVal, (m, n), nnzA)
             end
             return dA
         end
diff --git a/src/array.jl b/src/array.jl
index d576cdb..3225b5e 100644
--- a/src/array.jl
+++ b/src/array.jl
@@ -279,8 +279,9 @@ end
 
 ## interop with GPU arrays
 
-function Base.unsafe_convert(::Type{oneDeviceArray{T,N,AS.CrossWorkgroup}}, a::oneArray{T,N}) where {T,N}
-  oneDeviceArray{T,N,AS.CrossWorkgroup}(size(a), reinterpret(LLVMPtr{T,AS.CrossWorkgroup}, pointer(a)),
+function Base.unsafe_convert(::Type{oneDeviceArray{T, N, AS.CrossWorkgroup}}, a::oneArray{T, N}) where {T, N}
+    return oneDeviceArray{T, N, AS.CrossWorkgroup}(
+        size(a), reinterpret(LLVMPtr{T, AS.CrossWorkgroup}, pointer(a)),
                                 a.maxsize - a.offset*Base.elsize(a))
 end
 
diff --git a/src/broadcast.jl b/src/broadcast.jl
index 58a30dd..f9d38f6 100644
--- a/src/broadcast.jl
+++ b/src/broadcast.jl
@@ -11,9 +11,9 @@ BroadcastStyle(W::Type{<:oneWrappedArray{T, N}}) where {T, N} =
 # when we are dealing with different buffer styles, we cannot know
 # which one is better, so use shared memory
 BroadcastStyle(
-        ::oneArrayStyle{N, B1},
-        ::oneArrayStyle{N, B2},
-    ) where {N,B1,B2} =
+    ::oneArrayStyle{N, B1},
+    ::oneArrayStyle{N, B2},
+) where {N, B1, B2} =
     oneArrayStyle{N, oneL0.SharedBuffer}()
 
 # allocation of output arrays
diff --git a/src/compiler/execution.jl b/src/compiler/execution.jl
index 6503a9b..4900e77 100644
--- a/src/compiler/execution.jl
+++ b/src/compiler/execution.jl
@@ -88,7 +88,7 @@ Adapt.adapt_storage(to::KernelAdaptor, p::ZePtr{T}) where {T} = reinterpret(Ptr{
 
 # convert oneAPI host arrays to device arrays
 Adapt.adapt_storage(::KernelAdaptor, xs::oneArray{T,N}) where {T,N} =
-  Base.unsafe_convert(oneDeviceArray{T,N,AS.CrossWorkgroup}, xs)
+    Base.unsafe_convert(oneDeviceArray{T, N, AS.CrossWorkgroup}, xs)
 
 # Base.RefValue isn't GPU compatible, so provide a compatible alternative.
 # TODO: port improvements from CUDA.jl
diff --git a/src/indexing.jl b/src/indexing.jl
index d46c67b..13d1947 100644
--- a/src/indexing.jl
+++ b/src/indexing.jl
@@ -26,9 +26,9 @@ function Base.findall(bools::oneArray{Bool})
     ys = oneArray{I}(undef, n)
 
     if n > 0
-        kernel = @oneapi launch=false _ker!(ys, bools, indices)
+        kernel = @oneapi launch = false _ker!(ys, bools, indices)
         group_size = launch_configuration(kernel)
-        kernel(ys, bools, indices; items=group_size, groups=cld(length(bools), group_size))
+        kernel(ys, bools, indices; items = group_size, groups = cld(length(bools), group_size))
     end
     # unsafe_free!(indices)
 
diff --git a/test/device/intrinsics.jl b/test/device/intrinsics.jl
index 5e5605e..e1991fd 100644
--- a/test/device/intrinsics.jl
+++ b/test/device/intrinsics.jl
@@ -226,7 +226,7 @@ end
 
         s[t] = d[t]
         s2[t] = 2*d[t]
-        barrier(0)
+                barrier(0)
         d[t] = s[tr]
 
         return
@@ -252,7 +252,7 @@ end
 
             s[t] = d[t]
             s2[t] = d[t]
-            barrier(0)
+                    barrier(0)
             d[t] = s[tr]
 
             return
diff --git a/test/execution.jl b/test/execution.jl
index cd3db01..78f87d6 100644
--- a/test/execution.jl
+++ b/test/execution.jl
@@ -307,12 +307,12 @@ end
     @oneapi kernel(arr)
     @test Array(arr)[] == 1
 
-    function kernel2(ptr)
+        function kernel2(ptr)
         ptr[] = 2
         return
     end
 
-    @oneapi kernel2(arr)
+        @oneapi kernel2(arr)
     @test Array(arr)[] == 2
 end
 

[codecov]

Codecov Report

❌ Patch coverage is 91.50943% with 9 lines in your changes missing coverage. Please review.
✅ Project coverage is 79.21%. Comparing base (24110a8) to head (47c3ff7).

Files with missing lines	Patch %	Lines
lib/mkl/wrappers_sparse.jl	90.90%	5 Missing ⚠️
lib/mkl/wrappers_blas.jl	90.47%	4 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #534      +/-   ##
==========================================
+ Coverage   79.10%   79.21%   +0.10%     
==========================================
  Files          47       48       +1     
  Lines        3001     3021      +20     
==========================================
+ Hits         2374     2393      +19     
- Misses        627      628       +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.