Adapt to JuliaGPU/GPUArrays.jl#567.

[maleadt]

No description provided.

[github-actions]

Metal Benchmarks

Benchmark suite	Current: ac537f8	Previous: 15ac66d	Ratio
private array/construct	27889 ns	27827.428571428572 ns	1.00
private array/broadcast	462083 ns	465541 ns	0.99
private array/random/randn/Float32	814771 ns	831500 ns	0.98
private array/random/randn!/Float32	666167 ns	656625 ns	1.01
private array/random/rand!/Int64	568708 ns	548229 ns	1.04
private array/random/rand!/Float32	590458 ns	596000 ns	0.99
private array/random/rand/Int64	751375 ns	772125 ns	0.97
private array/random/rand/Float32	594499.5 ns	616292 ns	0.96
private array/copyto!/gpu_to_gpu	649959 ns	705667 ns	0.92
private array/copyto!/cpu_to_gpu	687354 ns	630625 ns	1.09
private array/copyto!/gpu_to_cpu	786625 ns	825292 ns	0.95
private array/accumulate/1d	1325583 ns	1304812.5 ns	1.02
private array/accumulate/2d	1389958 ns	1382792 ns	1.01
private array/iteration/findall/int	2076541.5 ns	2027166.5 ns	1.02
private array/iteration/findall/bool	1859166 ns	1816875 ns	1.02
private array/iteration/findfirst/int	1754187.5 ns	1688271 ns	1.04
private array/iteration/findfirst/bool	1640333 ns	1629458 ns	1.01
private array/iteration/scalar	3894708 ns	3825958 ns	1.02
private array/iteration/logical	3176375 ns	3194354.5 ns	0.99
private array/iteration/findmin/1d	1765167 ns	1736437.5 ns	1.02
private array/iteration/findmin/2d	1347687.5 ns	1346833 ns	1.00
private array/reductions/reduce/1d	1041250 ns	1045562 ns	1.00
private array/reductions/reduce/2d	645333 ns	647334 ns	1.00
private array/reductions/mapreduce/1d	1049333 ns	1051375 ns	1.00
private array/reductions/mapreduce/2d	656479 ns	656291 ns	1.00
private array/permutedims/4d	2681187.5 ns	2701896 ns	0.99
private array/permutedims/2d	1009583 ns	1005771 ns	1.00
private array/permutedims/3d	1559729 ns	1561083 ns	1.00
private array/copy	593729.5 ns	599500.5 ns	0.99
latency/precompile	5183655708 ns	5210869875 ns	0.99
latency/ttfp	6577854167 ns	6541068271 ns	1.01
latency/import	1127824729 ns	1140328708 ns	0.99
integration/metaldevrt	715875 ns	694000 ns	1.03
integration/byval/slices=1	1634854 ns	1586291.5 ns	1.03
integration/byval/slices=3	8445250 ns	10928917 ns	0.77
integration/byval/reference	1642334 ns	1622646 ns	1.01
integration/byval/slices=2	2573125 ns	2588083 ns	0.99
kernel/indexing	447895.5 ns	488771 ns	0.92
kernel/indexing_checked	455458 ns	478083 ns	0.95
kernel/launch	10104 ns	8125 ns	1.24
metal/synchronization/stream	14667 ns	14750 ns	0.99
metal/synchronization/context	14958 ns	15000 ns	1.00
shared array/construct	26566 ns	25173.583333333336 ns	1.06
shared array/broadcast	476167 ns	470833 ns	1.01
shared array/random/randn/Float32	788333.5 ns	785625.5 ns	1.00
shared array/random/randn!/Float32	663916 ns	661417 ns	1.00
shared array/random/rand!/Int64	553792 ns	576917 ns	0.96
shared array/random/rand!/Float32	589895.5 ns	612500 ns	0.96
shared array/random/rand/Int64	758166.5 ns	809208.5 ns	0.94
shared array/random/rand/Float32	621833 ns	625083 ns	0.99
shared array/copyto!/gpu_to_gpu	87375 ns	87167 ns	1.00
shared array/copyto!/cpu_to_gpu	90625 ns	88083 ns	1.03
shared array/copyto!/gpu_to_cpu	83042 ns	84167 ns	0.99
shared array/accumulate/1d	1351083 ns	1339709 ns	1.01
shared array/accumulate/2d	1393646 ns	1382229 ns	1.01
shared array/iteration/findall/int	1836937.5 ns	1775500 ns	1.03
shared array/iteration/findall/bool	1565792 ns	1593041.5 ns	0.98
shared array/iteration/findfirst/int	1394000 ns	1388583 ns	1.00
shared array/iteration/findfirst/bool	1356896 ns	1350167 ns	1.00
shared array/iteration/scalar	160083 ns	157125 ns	1.02
shared array/iteration/logical	2975708.5 ns	2964584 ns	1.00
shared array/iteration/findmin/1d	1450729 ns	1459458 ns	0.99
shared array/iteration/findmin/2d	1362083 ns	1367083 ns	1.00
shared array/reductions/reduce/1d	732333 ns	740209 ns	0.99
shared array/reductions/reduce/2d	662083 ns	664125 ns	1.00
shared array/reductions/mapreduce/1d	744687.5 ns	748291 ns	1.00
shared array/reductions/mapreduce/2d	664979 ns	662708.5 ns	1.00
shared array/permutedims/4d	2721938 ns	2702396 ns	1.01
shared array/permutedims/2d	1005958 ns	1011791.5 ns	0.99
shared array/permutedims/3d	1577000 ns	1577625 ns	1.00
shared array/copy	247541 ns	245125 ns	1.01

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