Fix #451 and a couple other fixes

[christiangnrd]

Closes #451

[github-actions]

Metal Benchmarks

Benchmark suite	Current: e4ae1ca	Previous: cd9e68a	Ratio
private array/construct	24642.416666666664 ns	26732.583333333336 ns	0.92
private array/broadcast	449458.5 ns	461834 ns	0.97
private array/random/randn/Float32	1008584 ns	1007750 ns	1.00
private array/random/randn!/Float32	640875 ns	633312.5 ns	1.01
private array/random/rand!/Int64	573791 ns	581542 ns	0.99
private array/random/rand!/Float32	600062.5 ns	592916.5 ns	1.01
private array/random/rand/Int64	855459 ns	870916 ns	0.98
private array/random/rand/Float32	868479.5 ns	977958 ns	0.89
private array/copyto!/gpu_to_gpu	485167 ns	486875 ns	1.00
private array/copyto!/cpu_to_gpu	541437.5 ns	720666.5 ns	0.75
private array/copyto!/gpu_to_cpu	495333.5 ns	576000 ns	0.86
private array/accumulate/1d	1442479 ns	1445354 ns	1.00
private array/accumulate/2d	1495458 ns	1477250 ns	1.01
private array/iteration/findall/int	2273000 ns	2238875 ns	1.02
private array/iteration/findall/bool	2061875 ns	2031625 ns	1.01
private array/iteration/findfirst/int	1670312.5 ns	1678291.5 ns	1.00
private array/iteration/findfirst/bool	1658395.5 ns	1671396 ns	0.99
private array/iteration/scalar	2383167 ns	2361083 ns	1.01
private array/iteration/logical	3447479 ns	3433396 ns	1.00
private array/iteration/findmin/1d	1766500 ns	1770396 ns	1.00
private array/iteration/findmin/2d	1363167 ns	1363729 ns	1.00
private array/reductions/reduce/1d	738000 ns	724645.5 ns	1.02
private array/reductions/reduce/2d	710208.5 ns	711916.5 ns	1.00
private array/reductions/mapreduce/1d	766583 ns	798667 ns	0.96
private array/reductions/mapreduce/2d	715562.5 ns	714500 ns	1.00
private array/permutedims/4d	943520.5 ns	946542 ns	1.00
private array/permutedims/2d	941333 ns	937833 ns	1.00
private array/permutedims/3d	1021375 ns	1007167 ns	1.01
private array/copy	762541.5 ns	893375 ns	0.85
latency/precompile	4411934458 ns	4410766833 ns	1.00
latency/ttfp	6903664750 ns	6887075395.5 ns	1.00
latency/import	726189854.5 ns	725464062.5 ns	1.00
integration/metaldevrt	756375 ns	755125 ns	1.00
integration/byval/slices=1	1621542 ns	1577625 ns	1.03
integration/byval/slices=3	8824396 ns	8895417 ns	0.99
integration/byval/reference	1622459 ns	1559209 ns	1.04
integration/byval/slices=2	2779250 ns	2635958 ns	1.05
kernel/indexing	447833 ns	462375 ns	0.97
kernel/indexing_checked	440250 ns	446500 ns	0.99
kernel/launch	11875 ns	11833 ns	1.00
metal/synchronization/stream	19458 ns	19542 ns	1.00
metal/synchronization/context	19875 ns	19833 ns	1.00
shared array/construct	23815.916666666664 ns	24125 ns	0.99
shared array/broadcast	470250 ns	467083.5 ns	1.01
shared array/random/randn/Float32	1009166.5 ns	1016896 ns	0.99
shared array/random/randn!/Float32	633417 ns	634250 ns	1.00
shared array/random/rand!/Int64	568959 ns	573167 ns	0.99
shared array/random/rand!/Float32	591417 ns	592958 ns	1.00
shared array/random/rand/Int64	852521 ns	865103.5 ns	0.99
shared array/random/rand/Float32	835771 ns	831271 ns	1.01
shared array/copyto!/gpu_to_gpu	643500 ns	678792 ns	0.95
shared array/copyto!/cpu_to_gpu	85083.5 ns	94625 ns	0.90
shared array/copyto!/gpu_to_cpu	84083 ns	84125 ns	1.00
shared array/accumulate/1d	1436833.5 ns	1446583 ns	0.99
shared array/accumulate/2d	1479750 ns	1483229.5 ns	1.00
shared array/iteration/findall/int	1975916 ns	1952167 ns	1.01
shared array/iteration/findall/bool	1780479.5 ns	1778083 ns	1.00
shared array/iteration/findfirst/int	1408937.5 ns	1415604 ns	1.00
shared array/iteration/findfirst/bool	1377667 ns	1376542 ns	1.00
shared array/iteration/scalar	190833 ns	192125 ns	0.99
shared array/iteration/logical	3211562.5 ns	3214833 ns	1.00
shared array/iteration/findmin/1d	1472604 ns	1477146 ns	1.00
shared array/iteration/findmin/2d	1379479 ns	1384958.5 ns	1.00
shared array/reductions/reduce/1d	652542 ns	644417 ns	1.01
shared array/reductions/reduce/2d	711708 ns	700500 ns	1.02
shared array/reductions/mapreduce/1d	616583 ns	673667 ns	0.92
shared array/reductions/mapreduce/2d	720000 ns	710479.5 ns	1.01
shared array/permutedims/4d	1072041 ns	935500 ns	1.15
shared array/permutedims/2d	943250 ns	944334 ns	1.00
shared array/permutedims/3d	1021375 ns	1023583 ns	1.00
shared array/copy	599625 ns	808959 ns	0.74

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