[maleadt]

GPUArrays 12.1.1 (JuliaGPU/GPUArrays.jl#429) shows GPU arrays as a call to their storage type, e.g. CuArray([1, 2]). This broke the CuSparseVector show test and a findmax doctest on every CUDA job and the docs build (main included, e.g. #3404). The new output evaluates to an equal array, so this updates the expectations, and requires 12.1.1 in the test environment so the old-dependencies job doesn't resolve 12.1.0.

[github-actions]

CUDA.jl Benchmarks

Details

Benchmark suite	Current: 387388c	Previous: 1b79f4f	Ratio
array/accumulate/Float32/1d	53366 ns	54029 ns	0.99
array/accumulate/Float32/dims=1	39105 ns	41851 ns	0.93
array/accumulate/Float32/dims=1L	82025 ns	82389 ns	1.00
array/accumulate/Float32/dims=2	48912 ns	49558 ns	0.99
array/accumulate/Float32/dims=2L	397120 ns	397688 ns	1.00
array/accumulate/Int64/1d	87810 ns	89543 ns	0.98
array/accumulate/Int64/dims=1	60211 ns	65047 ns	0.93
array/accumulate/Int64/dims=1L	140229 ns	140262 ns	1.00
array/accumulate/Int64/dims=2	67947 ns	68524 ns	0.99
array/accumulate/Int64/dims=2L	796228 ns	801148 ns	0.99
array/broadcast	10052 ns	10273 ns	0.98
array/broadcast launch	6501 ns	6735.6 ns	0.97
array/construct	897.3846153846154 ns	923.2758620689655 ns	0.97
array/copy	16897 ns	16830 ns	1.00
array/copyto!/cpu_to_gpu	206577 ns	208775 ns	0.99
array/copyto!/gpu_to_cpu	240077 ns	242835 ns	0.99
array/copyto!/gpu_to_gpu	8940 ns	8692 ns	1.03
array/iteration/findall/bool	46415 ns	46109 ns	1.01
array/iteration/findall/int	65318 ns	68182 ns	0.96
array/iteration/findfirst/bool	32512 ns	115342 ns	0.28
array/iteration/findfirst/int	32734 ns	113197 ns	0.29
array/iteration/findmin/1d	51096 ns	65849 ns	0.78
array/iteration/findmin/2d	50298 ns	66564 ns	0.76
array/iteration/logical	49258 ns	51642 ns	0.95
array/iteration/scalar	60971 ns	60297 ns	1.01
array/permutedims/2d	46791 ns	47023 ns	1.00
array/permutedims/3d	44043 ns	44336 ns	0.99
array/permutedims/4d	47297 ns	47618 ns	0.99
array/random/rand/Float32	11062 ns	10978 ns	1.01
array/random/rand/Int64	17572 ns	17663 ns	0.99
array/random/rand!/Float32	8278.666666666666 ns	8295 ns	1.00
array/random/rand!/Int64	15312 ns	15618 ns	0.98
array/random/randn/Float32	33936 ns	33770 ns	1.00
array/random/randn!/Float32	24923 ns	24898 ns	1.00
array/reductions/mapreduce/Float32/1d	32589 ns	48333 ns	0.67
array/reductions/mapreduce/Float32/dims=1	18463 ns	42319 ns	0.44
array/reductions/mapreduce/Float32/dims=1L	56442 ns	71774 ns	0.79
array/reductions/mapreduce/Float32/dims=2	26321 ns	42859 ns	0.61
array/reductions/mapreduce/Float32/dims=2L	56725 ns	72655 ns	0.78
array/reductions/mapreduce/Int64/1d	33311 ns	48949 ns	0.68
array/reductions/mapreduce/Int64/dims=1	20811 ns	45275 ns	0.46
array/reductions/mapreduce/Int64/dims=1L	95920 ns	111717 ns	0.86
array/reductions/mapreduce/Int64/dims=2	27034 ns	45401 ns	0.60
array/reductions/mapreduce/Int64/dims=2L	82477 ns	99464 ns	0.83
array/reductions/reduce/Float32/1d	32951 ns	48387 ns	0.68
array/reductions/reduce/Float32/dims=1	18807 ns	42567 ns	0.44
array/reductions/reduce/Float32/dims=1L	56979 ns	71973 ns	0.79
array/reductions/reduce/Float32/dims=2	26592 ns	44021 ns	0.60
array/reductions/reduce/Float32/dims=2L	56460 ns	73036 ns	0.77
array/reductions/reduce/Int64/1d	33649 ns	49228 ns	0.68
array/reductions/reduce/Int64/dims=1	21281 ns	44663 ns	0.48
array/reductions/reduce/Int64/dims=1L	96236 ns	112074 ns	0.86
array/reductions/reduce/Int64/dims=2	27031 ns	45274 ns	0.60
array/reductions/reduce/Int64/dims=2L	82550 ns	99445 ns	0.83
array/reverse/1d	19445 ns	23449 ns	0.83
array/reverse/1dL	73255 ns	129199 ns	0.57
array/reverse/1dL_inplace	70173 ns	70208 ns	1.00
array/reverse/1d_inplace	12043 ns	12124 ns	0.99
array/reverse/2d	22020 ns	22253 ns	0.99
array/reverse/2dL	80452 ns	80779 ns	1.00
array/reverse/2dL_inplace	70701 ns	70674 ns	1.00
array/reverse/2d_inplace	14436 ns	14685 ns	0.98
array/sorting/1d	748846 ns	750547 ns	1.00
array/sorting/2d	193727 ns	194284 ns	1.00
array/sorting/by	927153 ns	928102 ns	1.00
cuda/graph/capture	34770 ns	36681 ns	0.95
cuda/graph/eager	32159 ns	33446 ns	0.96
cuda/graph/launch	13798 ns	13961 ns	0.99
cuda/synchronization/context/auto	2325.5555555555557 ns	2393.3333333333335 ns	0.97
cuda/synchronization/context/blocking	890.3541666666666 ns	883.8979591836735 ns	1.01
cuda/synchronization/context/nonblocking	9902 ns	8554 ns	1.16
cuda/synchronization/stream/auto	873.421052631579 ns	800.952380952381 ns	1.09
cuda/synchronization/stream/blocking	766 ns	751.4867256637168 ns	1.02
cuda/synchronization/stream/nonblocking	7785.666666666667 ns	7901 ns	0.99
integration/byval/reference	147457 ns	147819 ns	1.00
integration/byval/slices=1	148678 ns	148798 ns	1.00
integration/byval/slices=2	290235 ns	290480 ns	1.00
integration/byval/slices=3	432231 ns	432462 ns	1.00
integration/cudadevrt	104891 ns	105075 ns	1.00
integration/volumerhs	9608621 ns	9610241 ns	1.00
kernel/indexing	11294 ns	11469 ns	0.98
kernel/indexing_checked	11712 ns	12109 ns	0.97
kernel/launch	2271.6666666666665 ns	2304.3333333333335 ns	0.99
kernel/occupancy	677.4640522875817 ns	675.6887417218543 ns	1.00
kernel/rand	14682 ns	15003 ns	0.98
latency/import	4784677907 ns	4752896759 ns	1.01
latency/precompile	5428799973 ns	5388281881 ns	1.01
latency/ttfp	5446270279 ns	5254389535 ns	1.04

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