Adapt to minver in ObjectiveC.jl

[christiangnrd]

Mostly to show JuliaInterop/ObjectiveC.jl#47 in action.

To try it out, make just to dev the ObjectiveC branch, then:

julia> using Metal; dev = device(); ndarr = MPS.MPSNDArray(dev, 3); propertynames(ndarr)
13-element Vector{Symbol}:
 :numberOfDimensions
 :dataTypeSize
 :dataType
 :device
 :label
 :parent
 :descriptor
 :userBuffer
 :resourceSize
 :retainCount
 :description
 :debugDescription
 :hash

If you change "15" to "16", you get:

julia> using Metal; dev = device(); ndarr = MPS.MPSNDArray(dev, 3); propertynames(ndarr)
10-element Vector{Symbol}:
 :numberOfDimensions
 :dataTypeSize
 :dataType
 :device
 :label
 :parent
 :retainCount
 :description
 :debugDescription
 :hash

[github-actions]

Metal Benchmarks

Benchmark suite	Current: fe3a68a	Previous: 6a760a6	Ratio
private array/construct	27916.714285714286 ns	27270.785714285714 ns	1.02
private array/broadcast	463166 ns	460209 ns	1.01
private array/random/randn/Float32	757333 ns	804875 ns	0.94
private array/random/randn!/Float32	662583 ns	646333 ns	1.03
private array/random/rand!/Int64	563895.5 ns	548916 ns	1.03
private array/random/rand!/Float32	592083.5 ns	587375 ns	1.01
private array/random/rand/Int64	738958.5 ns	716083.5 ns	1.03
private array/random/rand/Float32	617000 ns	615791.5 ns	1.00
private array/copyto!/gpu_to_gpu	711542 ns	677125 ns	1.05
private array/copyto!/cpu_to_gpu	649875 ns	640688 ns	1.01
private array/copyto!/gpu_to_cpu	790792 ns	817625 ns	0.97
private array/accumulate/1d	1322500 ns	1329687.5 ns	0.99
private array/accumulate/2d	1384270.5 ns	1382229 ns	1.00
private array/iteration/findall/int	2068333.5 ns	2073708 ns	1.00
private array/iteration/findall/bool	1822937.5 ns	1799041 ns	1.01
private array/iteration/findfirst/int	1727875 ns	1688292 ns	1.02
private array/iteration/findfirst/bool	1666375 ns	1650020.5 ns	1.01
private array/iteration/scalar	3845792 ns	3252542 ns	1.18
private array/iteration/logical	3187000 ns	3147375 ns	1.01
private array/iteration/findmin/1d	1752000 ns	1736042 ns	1.01
private array/iteration/findmin/2d	1353167 ns	1348917 ns	1.00
private array/reductions/reduce/1d	1045417 ns	1029542 ns	1.02
private array/reductions/reduce/2d	662104.5 ns	650292 ns	1.02
private array/reductions/mapreduce/1d	1045583.5 ns	1025917 ns	1.02
private array/reductions/mapreduce/2d	658083 ns	657229.5 ns	1.00
private array/permutedims/4d	2522375 ns	2553708 ns	0.99
private array/permutedims/2d	1015667 ns	1027750 ns	0.99
private array/permutedims/3d	1591062.5 ns	1585916 ns	1.00
private array/copy	601000 ns	580417 ns	1.04
latency/precompile	5864734437.5 ns	5847134584 ns	1.00
latency/ttfp	6623113124.5 ns	6545482667 ns	1.01
latency/import	1171809750 ns	1169724375 ns	1.00
integration/metaldevrt	704250 ns	713125 ns	0.99
integration/byval/slices=1	1575250 ns	1580770.5 ns	1.00
integration/byval/slices=3	11669708 ns	9774042 ns	1.19
integration/byval/reference	1559667 ns	1598000 ns	0.98
integration/byval/slices=2	2649104.5 ns	2571895.5 ns	1.03
kernel/indexing	453875 ns	457542 ns	0.99
kernel/indexing_checked	452417 ns	458645.5 ns	0.99
kernel/launch	8084 ns	8125 ns	0.99
metal/synchronization/stream	14750 ns	14209 ns	1.04
metal/synchronization/context	14875 ns	15000 ns	0.99
shared array/construct	27649.25 ns	25166.714285714286 ns	1.10
shared array/broadcast	461708 ns	469917 ns	0.98
shared array/random/randn/Float32	818250 ns	825666 ns	0.99
shared array/random/randn!/Float32	681458 ns	616625 ns	1.11
shared array/random/rand!/Int64	580792 ns	547708 ns	1.06
shared array/random/rand!/Float32	602708 ns	591458 ns	1.02
shared array/random/rand/Int64	787875 ns	734166.5 ns	1.07
shared array/random/rand/Float32	619792 ns	610312 ns	1.02
shared array/copyto!/gpu_to_gpu	87291 ns	88000 ns	0.99
shared array/copyto!/cpu_to_gpu	88250 ns	86291 ns	1.02
shared array/copyto!/gpu_to_cpu	83625 ns	77959 ns	1.07
shared array/accumulate/1d	1345375 ns	1336542 ns	1.01
shared array/accumulate/2d	1386041 ns	1384250 ns	1.00
shared array/iteration/findall/int	1768583 ns	1765020.5 ns	1.00
shared array/iteration/findall/bool	1572979 ns	1558812 ns	1.01
shared array/iteration/findfirst/int	1415334 ns	1396125.5 ns	1.01
shared array/iteration/findfirst/bool	1362250 ns	1360145.5 ns	1.00
shared array/iteration/scalar	158708 ns	153334 ns	1.04
shared array/iteration/logical	2969687.5 ns	2940000 ns	1.01
shared array/iteration/findmin/1d	1463437 ns	1442291 ns	1.01
shared array/iteration/findmin/2d	1372854 ns	1358562.5 ns	1.01
shared array/reductions/reduce/1d	744792 ns	723750 ns	1.03
shared array/reductions/reduce/2d	662895.5 ns	654792 ns	1.01
shared array/reductions/mapreduce/1d	741875 ns	728687 ns	1.02
shared array/reductions/mapreduce/2d	665542 ns	660125 ns	1.01
shared array/permutedims/4d	2537125 ns	2522521 ns	1.01
shared array/permutedims/2d	1032083 ns	1019375 ns	1.01
shared array/permutedims/3d	1589625 ns	1579917 ns	1.01
shared array/copy	242916 ns	233417 ns	1.04

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