Even more CUSPARSE tests

[kshyatt]

We should turn off coverage for the device-side file. Added some simple tests for unsafe_free!, happy to improve them if there's a better idea.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 82.59%. Comparing base (2540087) to head (49edc86).
Report is 2 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2682      +/-   ##
==========================================
+ Coverage   82.08%   82.59%   +0.50%     
==========================================
  Files         154      153       -1     
  Lines       13661    13606      -55     
==========================================
+ Hits        11214    11238      +24     
+ Misses       2447     2368      -79     

☔ 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.

[github-actions]

CUDA.jl Benchmarks

Benchmark suite	Current: 49edc86	Previous: 2540087	Ratio
latency/precompile	46191914324 ns	46159355585.5 ns	1.00
latency/ttfp	7048535341 ns	6957794099 ns	1.01
latency/import	3680835625 ns	3631949691 ns	1.01
integration/volumerhs	9622492 ns	9611328 ns	1.00
integration/byval/slices=1	147029 ns	146783 ns	1.00
integration/byval/slices=3	425204 ns	425400 ns	1.00
integration/byval/reference	144986 ns	145027 ns	1.00
integration/byval/slices=2	286029 ns	286289 ns	1.00
integration/cudadevrt	103432 ns	103478 ns	1.00
kernel/indexing	14059 ns	14116.5 ns	1.00
kernel/indexing_checked	14809 ns	14717 ns	1.01
kernel/occupancy	635.5411764705882 ns	631.1345029239766 ns	1.01
kernel/launch	2024.1 ns	2048.6 ns	0.99
kernel/rand	17910 ns	15306 ns	1.17
array/reverse/1d	19466 ns	19766 ns	0.98
array/reverse/2d	23038 ns	23320 ns	0.99
array/reverse/1d_inplace	10000 ns	10115 ns	0.99
array/reverse/2d_inplace	11642 ns	11769 ns	0.99
array/copy	20904 ns	21157 ns	0.99
array/iteration/findall/int	157174 ns	159462 ns	0.99
array/iteration/findall/bool	138272.5 ns	139925 ns	0.99
array/iteration/findfirst/int	152707.5 ns	154378.5 ns	0.99
array/iteration/findfirst/bool	154040 ns	155307 ns	0.99
array/iteration/scalar	72483 ns	74069 ns	0.98
array/iteration/logical	212005 ns	217157 ns	0.98
array/iteration/findmin/1d	40701 ns	41948 ns	0.97
array/iteration/findmin/2d	93323 ns	94228.5 ns	0.99
array/reductions/reduce/1d	35696 ns	36315 ns	0.98
array/reductions/reduce/2d	40565 ns	40828 ns	0.99
array/reductions/mapreduce/1d	33024 ns	33656 ns	0.98
array/reductions/mapreduce/2d	40709 ns	51092 ns	0.80
array/broadcast	20641 ns	21055 ns	0.98
array/copyto!/gpu_to_gpu	13692 ns	11958 ns	1.15
array/copyto!/cpu_to_gpu	208318 ns	211255 ns	0.99
array/copyto!/gpu_to_cpu	243872 ns	244927 ns	1.00
array/accumulate/1d	108093 ns	109111 ns	0.99
array/accumulate/2d	79684 ns	80441 ns	0.99
array/construct	1256.6 ns	1271.2 ns	0.99
array/random/randn/Float32	42847.5 ns	47305 ns	0.91
array/random/randn!/Float32	26167 ns	26683 ns	0.98
array/random/rand!/Int64	27091 ns	27072 ns	1.00
array/random/rand!/Float32	8865.666666666666 ns	8803.333333333334 ns	1.01
array/random/rand/Int64	29636 ns	29816 ns	0.99
array/random/rand/Float32	12962 ns	13176 ns	0.98
array/permutedims/4d	61115 ns	61379.5 ns	1.00
array/permutedims/2d	55102.5 ns	55887 ns	0.99
array/permutedims/3d	55745.5 ns	56426.5 ns	0.99
array/sorting/1d	2775958 ns	2766596.5 ns	1.00
array/sorting/by	3366877.5 ns	3370290 ns	1.00
array/sorting/2d	1084115 ns	1085614.5 ns	1.00
cuda/synchronization/stream/auto	1024 ns	1027.9 ns	1.00
cuda/synchronization/stream/nonblocking	6480.8 ns	6419.6 ns	1.01
cuda/synchronization/stream/blocking	802.2083333333334 ns	800.59 ns	1.00
cuda/synchronization/context/auto	1146.2 ns	1152.4 ns	0.99
cuda/synchronization/context/nonblocking	6601 ns	6599.2 ns	1.00
cuda/synchronization/context/blocking	893.3846153846154 ns	899.7446808510638 ns	0.99

This comment was automatically generated by workflow using github-action-benchmark.

[github-actions]

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

Click here to view the suggested changes.

diff --git a/lib/cusparse/array.jl b/lib/cusparse/array.jl
index 2f411eebd..3c598b66b 100644
--- a/lib/cusparse/array.jl
+++ b/lib/cusparse/array.jl
@@ -490,7 +490,8 @@ CuSparseVector{T}(Mat::SparseMatrixCSC) where {T} =
         throw(ArgumentError("The input argument must have a single column"))
 CuSparseMatrixCSC{T}(Vec::SparseVector) where {T} =
     CuSparseMatrixCSC{T}(CuVector{Cint}([1]), CuVector{Cint}(Vec.nzind),
-                         CuVector{T}(Vec.nzval), (length(Vec), 1))
+    CuVector{T}(Vec.nzval), (length(Vec), 1)
+)
 CuSparseMatrixCSC{T}(Mat::SparseMatrixCSC) where {T} =
     CuSparseMatrixCSC{T}(CuVector{Cint}(Mat.colptr), CuVector{Cint}(Mat.rowval),
                          CuVector{T}(Mat.nzval), size(Mat))
diff --git a/test/libraries/cusparse.jl b/test/libraries/cusparse.jl
index 843d49980..c0c4e878c 100644
--- a/test/libraries/cusparse.jl
+++ b/test/libraries/cusparse.jl
@@ -22,7 +22,7 @@ blockdim = 5
     @test ndims(d_x)  == 1
     dense_d_x = CuVector(x)
     CUDA.@allowscalar begin
-        @test sprint(show, d_x) == replace(sprint(show, x), "SparseVector{Float64, Int64}"=>"CUDA.CUSPARSE.CuSparseVector{Float64, Int32}", "sparsevec(["=>"sparsevec(Int32[")
+        @test sprint(show, d_x) == replace(sprint(show, x), "SparseVector{Float64, Int64}" => "CUDA.CUSPARSE.CuSparseVector{Float64, Int32}", "sparsevec([" => "sparsevec(Int32[")
         @test Array(d_x[:])        == x[:]
         @test d_x[firstindex(d_x)] == x[firstindex(x)]
         @test d_x[div(end, 2)]     == x[div(end, 2)]
@@ -39,7 +39,7 @@ blockdim = 5
     @test nnz(d_x)    == length(nonzeros(d_x))
     d_y = copy(d_x)
     CUDA.unsafe_free!(d_y)
-    x = sprand(m,0.2)
+    x = sprand(m, 0.2)
     d_x = CuSparseMatrixCSC{Float64}(x)
     @test size(d_x) == (m, 1)
     x = sprand(m,n,0.2)

[maleadt]

CI failure looks related.