Simplify sparse type mappings

[kshyatt]

Needs JuliaGPU/GPUArrays.jl#700 first

cc @gdalle

[github-actions]

CUDA.jl Benchmarks

Details

Benchmark suite	Current: 5dd9f87	Previous: 340127b	Ratio
array/accumulate/Float32/1d	100754 ns	101458 ns	0.99
array/accumulate/Float32/dims=1	76473 ns	77354.5 ns	0.99
array/accumulate/Float32/dims=1L	1585449 ns	1586530.5 ns	1.00
array/accumulate/Float32/dims=2	143371 ns	144422 ns	0.99
array/accumulate/Float32/dims=2L	657279 ns	658433.5 ns	1.00
array/accumulate/Int64/1d	118263 ns	118510 ns	1.00
array/accumulate/Int64/dims=1	79581 ns	79806 ns	1.00
array/accumulate/Int64/dims=1L	1706091 ns	1705339 ns	1.00
array/accumulate/Int64/dims=2	156925 ns	156169.5 ns	1.00
array/accumulate/Int64/dims=2L	961168 ns	961890 ns	1.00
array/broadcast	20513 ns	20634 ns	0.99
array/construct	1274.6 ns	1265.4 ns	1.01
array/copy	18269 ns	18079 ns	1.01
array/copyto!/cpu_to_gpu	214123 ns	216989 ns	0.99
array/copyto!/gpu_to_cpu	280436 ns	284915 ns	0.98
array/copyto!/gpu_to_gpu	10977 ns	10779 ns	1.02
array/iteration/findall/bool	134196 ns	134702.5 ns	1.00
array/iteration/findall/int	148851 ns	150167 ns	0.99
array/iteration/findfirst/bool	80843 ns	81791 ns	0.99
array/iteration/findfirst/int	83434 ns	83856 ns	0.99
array/iteration/findmin/1d	86703.5 ns	86603.5 ns	1.00
array/iteration/findmin/2d	116663.5 ns	117335 ns	0.99
array/iteration/logical	195739 ns	198412.5 ns	0.99
array/iteration/scalar	67797 ns	69274.5 ns	0.98
array/permutedims/2d	51765 ns	52565 ns	0.98
array/permutedims/3d	52061 ns	52656 ns	0.99
array/permutedims/4d	51013 ns	51433.5 ns	0.99
array/random/rand/Float32	12888 ns	13120 ns	0.98
array/random/rand/Int64	24944 ns	24957 ns	1.00
array/random/rand!/Float32	8401.666666666666 ns	9858.333333333334 ns	0.85
array/random/rand!/Int64	21701 ns	21835 ns	0.99
array/random/randn/Float32	36602 ns	43515.5 ns	0.84
array/random/randn!/Float32	30844 ns	30949.5 ns	1.00
array/reductions/mapreduce/Float32/1d	34290.5 ns	34651 ns	0.99
array/reductions/mapreduce/Float32/dims=1	49209.5 ns	44540.5 ns	1.10
array/reductions/mapreduce/Float32/dims=1L	51219 ns	51138 ns	1.00
array/reductions/mapreduce/Float32/dims=2	56416 ns	56567 ns	1.00
array/reductions/mapreduce/Float32/dims=2L	69240 ns	69274 ns	1.00
array/reductions/mapreduce/Int64/1d	42241 ns	43548.5 ns	0.97
array/reductions/mapreduce/Int64/dims=1	50891.5 ns	43771.5 ns	1.16
array/reductions/mapreduce/Int64/dims=1L	87167 ns	87029 ns	1.00
array/reductions/mapreduce/Int64/dims=2	59275 ns	59317 ns	1.00
array/reductions/mapreduce/Int64/dims=2L	84506 ns	84771 ns	1.00
array/reductions/reduce/Float32/1d	34558 ns	35127 ns	0.98
array/reductions/reduce/Float32/dims=1	39512 ns	42032 ns	0.94
array/reductions/reduce/Float32/dims=1L	51243 ns	51445 ns	1.00
array/reductions/reduce/Float32/dims=2	56338 ns	57148 ns	0.99
array/reductions/reduce/Float32/dims=2L	69769 ns	70183 ns	0.99
array/reductions/reduce/Int64/1d	42569 ns	43251 ns	0.98
array/reductions/reduce/Int64/dims=1	41961 ns	51781 ns	0.81
array/reductions/reduce/Int64/dims=1L	86985 ns	87117 ns	1.00
array/reductions/reduce/Int64/dims=2	59408 ns	59507 ns	1.00
array/reductions/reduce/Int64/dims=2L	84411 ns	84592 ns	1.00
array/reverse/1d	17512 ns	17712 ns	0.99
array/reverse/1dL	68271 ns	68269 ns	1.00
array/reverse/1dL_inplace	65675 ns	65689 ns	1.00
array/reverse/1d_inplace	10248 ns	10212.666666666666 ns	1.00
array/reverse/2d	20802 ns	20819 ns	1.00
array/reverse/2dL	72786 ns	72976 ns	1.00
array/reverse/2dL_inplace	65736 ns	65812 ns	1.00
array/reverse/2d_inplace	10221 ns	10867 ns	0.94
array/sorting/1d	2734313.5 ns	2735648 ns	1.00
array/sorting/2d	1067889 ns	1069639 ns	1.00
array/sorting/by	3303732 ns	3305008 ns	1.00
cuda/synchronization/context/auto	1192.3 ns	1149 ns	1.04
cuda/synchronization/context/blocking	968.4827586206897 ns	961.56 ns	1.01
cuda/synchronization/context/nonblocking	7872.1 ns	6843.4 ns	1.15
cuda/synchronization/stream/auto	997.3 ns	1040.15 ns	0.96
cuda/synchronization/stream/blocking	827.7010309278351 ns	834.2666666666667 ns	0.99
cuda/synchronization/stream/nonblocking	7482 ns	7317.4 ns	1.02
integration/byval/reference	143800.5 ns	143691 ns	1.00
integration/byval/slices=1	145773 ns	145548 ns	1.00
integration/byval/slices=2	284740 ns	284456 ns	1.00
integration/byval/slices=3	423037 ns	423048 ns	1.00
integration/cudadevrt	102329 ns	102314 ns	1.00
integration/volumerhs	23439655.5 ns	23431006.5 ns	1.00
kernel/indexing	13295 ns	13281.5 ns	1.00
kernel/indexing_checked	14021 ns	13966 ns	1.00
kernel/launch	2028.9 ns	2168.222222222222 ns	0.94
kernel/occupancy	667.2967741935483 ns	669.85625 ns	1.00
kernel/rand	15019 ns	14425 ns	1.04
latency/import	3825173202 ns	3813842835.5 ns	1.00
latency/precompile	4577693352.5 ns	4579317015 ns	1.00
latency/ttfp	4418922320.5 ns	4395787908 ns	1.01

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

[gdalle]

Looks like it's time to tag a release of GPUArrays?