Added Base.similar methods for CuSparseMatrixCOO and BSR

[rainerrodrigues]

This PR adds the missing Base.similar methods for CuSparseMatrixCOO and CuSparseMatrixBSR, allowing them to fallback gracefully without converting to dense CPU arrays.

Fixes #3061
Fixes #3055

[kshyatt]

Also, can some tests be added?

[github-actions]

CUDA.jl Benchmarks

Details

Benchmark suite	Current: 07cb674	Previous: fdb0f83	Ratio
array/accumulate/Float32/1d	99956 ns	98753 ns	1.01
array/accumulate/Float32/dims=1	76223 ns	74838 ns	1.02
array/accumulate/Float32/dims=1L	1594955 ns	1595600 ns	1.00
array/accumulate/Float32/dims=2	141352 ns	139337 ns	1.01
array/accumulate/Float32/dims=2L	653944 ns	653321 ns	1.00
array/accumulate/Int64/1d	119204 ns	118217 ns	1.01
array/accumulate/Int64/dims=1	80206 ns	79158 ns	1.01
array/accumulate/Int64/dims=1L	1710205 ns	1709186 ns	1.00
array/accumulate/Int64/dims=2	154209 ns	153413 ns	1.01
array/accumulate/Int64/dims=2L	960254 ns	959320 ns	1.00
array/broadcast	18913 ns	18499 ns	1.02
array/construct	1199 ns	1232 ns	0.97
array/copy	16928 ns	16662 ns	1.02
array/copyto!/cpu_to_gpu	216158 ns	210966 ns	1.02
array/copyto!/gpu_to_cpu	280211 ns	278113 ns	1.01
array/copyto!/gpu_to_gpu	10403 ns	10251.333333333334 ns	1.01
array/iteration/findall/bool	134493 ns	132507 ns	1.01
array/iteration/findall/int	148906 ns	147140 ns	1.01
array/iteration/findfirst/bool	70933 ns	70124 ns	1.01
array/iteration/findfirst/int	72031 ns	71302 ns	1.01
array/iteration/findmin/1d	69673 ns	67538 ns	1.03
array/iteration/findmin/2d	101491 ns	101327 ns	1.00
array/iteration/logical	195763 ns	192237 ns	1.02
array/iteration/scalar	65862 ns	64116 ns	1.03
array/permutedims/2d	50412 ns	49526 ns	1.02
array/permutedims/3d	51182 ns	50756 ns	1.01
array/permutedims/4d	51456 ns	50974 ns	1.01
array/random/rand/Float32	12310 ns	11658 ns	1.06
array/random/rand/Int64	24741 ns	22560 ns	1.10
array/random/rand!/Float32	8276.333333333334 ns	7849.333333333333 ns	1.05
array/random/rand!/Int64	19746 ns	17870 ns	1.10
array/random/randn/Float32	35552 ns	35106 ns	1.01
array/random/randn!/Float32	23934 ns	23912 ns	1.00
array/reductions/mapreduce/Float32/1d	34001 ns	33612 ns	1.01
array/reductions/mapreduce/Float32/dims=1	39016 ns	38308 ns	1.02
array/reductions/mapreduce/Float32/dims=1L	50570 ns	50124 ns	1.01
array/reductions/mapreduce/Float32/dims=2	56113 ns	55421 ns	1.01
array/reductions/mapreduce/Float32/dims=2L	67503 ns	67539 ns	1.00
array/reductions/mapreduce/Int64/1d	39984 ns	39433 ns	1.01
array/reductions/mapreduce/Int64/dims=1	41545 ns	41004 ns	1.01
array/reductions/mapreduce/Int64/dims=1L	86823 ns	86427 ns	1.00
array/reductions/mapreduce/Int64/dims=2	58286 ns	57708 ns	1.01
array/reductions/mapreduce/Int64/dims=2L	83536 ns	82911 ns	1.01
array/reductions/reduce/Float32/1d	34274 ns	33312 ns	1.03
array/reductions/reduce/Float32/dims=1	38845 ns	38395 ns	1.01
array/reductions/reduce/Float32/dims=1L	50683 ns	50133 ns	1.01
array/reductions/reduce/Float32/dims=2	56245 ns	55712 ns	1.01
array/reductions/reduce/Float32/dims=2L	69362 ns	68679 ns	1.01
array/reductions/reduce/Int64/1d	40309 ns	39077 ns	1.03
array/reductions/reduce/Int64/dims=1	41975 ns	40942 ns	1.03
array/reductions/reduce/Int64/dims=1L	86880 ns	86287 ns	1.01
array/reductions/reduce/Int64/dims=2	58107 ns	57687 ns	1.01
array/reductions/reduce/Int64/dims=2L	83473 ns	82800 ns	1.01
array/reverse/1d	17322 ns	16733 ns	1.04
array/reverse/1dL	68140 ns	67771 ns	1.01
array/reverse/1dL_inplace	65505 ns	65314 ns	1.00
array/reverse/1d_inplace	8840.666666666666 ns	8293.666666666666 ns	1.07
array/reverse/2d	20879 ns	20003 ns	1.04
array/reverse/2dL	72366 ns	71847 ns	1.01
array/reverse/2dL_inplace	65251 ns	65036 ns	1.00
array/reverse/2d_inplace	9928 ns	9632 ns	1.03
array/sorting/1d	2660346 ns	2656101 ns	1.00
array/sorting/2d	1041945 ns	1038978 ns	1.00
array/sorting/by	3195808 ns	3193388 ns	1.00
cuda/synchronization/context/auto	1149.2 ns	1144.2 ns	1.00
cuda/synchronization/context/blocking	915.4848484848485 ns	945.2307692307693 ns	0.97
cuda/synchronization/context/nonblocking	6304.2 ns	5938.8 ns	1.06
cuda/synchronization/stream/auto	1010.3 ns	1002.3 ns	1.01
cuda/synchronization/stream/blocking	806.2022471910112 ns	835.5142857142857 ns	0.96
cuda/synchronization/stream/nonblocking	6130.5 ns	5969.4 ns	1.03
integration/byval/reference	143376 ns	143209 ns	1.00
integration/byval/slices=1	145681 ns	145483 ns	1.00
integration/byval/slices=2	284497 ns	283679 ns	1.00
integration/byval/slices=3	422437 ns	422104 ns	1.00
integration/cudadevrt	101862 ns	101601 ns	1.00
integration/volumerhs	9089170 ns	9087846 ns	1.00
kernel/indexing	13015 ns	12529 ns	1.04
kernel/indexing_checked	13886 ns	13442 ns	1.03
kernel/launch	2198.222222222222 ns	2078.777777777778 ns	1.06
kernel/occupancy	730.6590909090909 ns	762.8947368421053 ns	0.96
kernel/rand	14273 ns	14692 ns	0.97
latency/import	3880740734 ns	3855139894 ns	1.01
latency/precompile	4640823930 ns	4639622735 ns	1.00
latency/ttfp	4546678518 ns	4510104327 ns	1.01

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

[rainerrodrigues]

@kshyatt Hi, can you check if this is suitable and extensive enough for testing?

[maleadt]

Same as #3119, you seem to have many unrelated changes in here that cause CI failures.

[codecov]

Codecov Report

❌ Patch coverage is 0% with 6 lines in your changes missing coverage. Please review.
✅ Project coverage is 16.86%. Comparing base (642cf8d) to head (07cb674).
⚠️ Report is 5 commits behind head on main.

Files with missing lines	Patch %	Lines
lib/cusparse/src/array.jl	0.00%	6 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3114      +/-   ##
==========================================
+ Coverage   16.32%   16.86%   +0.53%     
==========================================
  Files         124      124              
  Lines        9875     9881       +6     
==========================================
+ Hits         1612     1666      +54     
+ Misses       8263     8215      -48     

☔ View full report in Codecov by Harness.
📢 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.