Support Julia 1.13 with fix for @device_functions macro

[KSepetanc]

Closes #3019.

@eschnett asked me to create a new duplicate PR #3020 of his, but with fix for macro @device_functions. He couldn't test if the fix works as I made PR on his fork that does not have CI infrastructure.

[eschnett]

Well, I didn't really ask for a duplicate PR. I suggested to either merge them into CUDA.jl as two separate, sequential PRs, or – if you want to merge them as a single PR into CUDA.jl – create such a PR. I don't care either way; using two separate PRs seems simpler, but I leave that choice up to you.

[KSepetanc]

The way I see it, this is the second option, i.e. single PR with both changes.

@device_functions macro is broken due to changes in 1.13 so it makes sense to include it in PR for 1.13 support.

[codecov]

Codecov Report

❌ Patch coverage is 50.00000% with 3 lines in your changes missing coverage. Please review.
✅ Project coverage is 89.35%. Comparing base (7a27d77) to head (aa08bb6).

Files with missing lines	Patch %	Lines
lib/nvml/NVML.jl	50.00%	3 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #3031      +/-   ##
==========================================
- Coverage   89.46%   89.35%   -0.12%     
==========================================
  Files         148      148              
  Lines       13047    13044       -3     
==========================================
- Hits        11673    11655      -18     
- Misses       1374     1389      +15     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[github-actions]

CUDA.jl Benchmarks

Details

Benchmark suite	Current: b1e9b57	Previous: 1810b7a	Ratio
latency/precompile	44328349517 ns	44300180944.5 ns	1.00
latency/ttfp	13141566320 ns	13138137112 ns	1.00
latency/import	3768894174.5 ns	3757487166.5 ns	1.00
integration/volumerhs	9442367.5 ns	9441754.5 ns	1.00
integration/byval/slices=1	145610 ns	145846 ns	1.00
integration/byval/slices=3	422716 ns	423265 ns	1.00
integration/byval/reference	143823.5 ns	143916 ns	1.00
integration/byval/slices=2	284142 ns	284641 ns	1.00
integration/cudadevrt	102447 ns	102633 ns	1.00
kernel/indexing	13360 ns	13466 ns	0.99
kernel/indexing_checked	14152 ns	13982 ns	1.01
kernel/occupancy	649.422619047619 ns	699.625850340136 ns	0.93
kernel/launch	2059.9 ns	2067.8 ns	1.00
kernel/rand	16211 ns	16244 ns	1.00
array/reverse/1d	18777 ns	18605 ns	1.01
array/reverse/2dL_inplace	66066 ns	66133 ns	1.00
array/reverse/1dL	69078 ns	68870 ns	1.00
array/reverse/2d	20795 ns	20781 ns	1.00
array/reverse/1d_inplace	10525.166666666668 ns	10493.666666666666 ns	1.00
array/reverse/2d_inplace	10614 ns	10765 ns	0.99
array/reverse/2dL	72846 ns	72777.5 ns	1.00
array/reverse/1dL_inplace	66097 ns	66166 ns	1.00
array/copy	18360.5 ns	18321 ns	1.00
array/iteration/findall/int	145080 ns	145251 ns	1.00
array/iteration/findall/bool	130258 ns	130303 ns	1.00
array/iteration/findfirst/int	82889 ns	83996 ns	0.99
array/iteration/findfirst/bool	80606 ns	81209 ns	0.99
array/iteration/scalar	66588 ns	64953 ns	1.03
array/iteration/logical	194180.5 ns	197334 ns	0.98
array/iteration/findmin/1d	83560.5 ns	85667.5 ns	0.98
array/iteration/findmin/2d	116518 ns	117130 ns	0.99
array/reductions/reduce/Int64/1d	39034 ns	38913 ns	1.00
array/reductions/reduce/Int64/dims=1	41876.5 ns	41855 ns	1.00
array/reductions/reduce/Int64/dims=2	58784 ns	59043 ns	1.00
array/reductions/reduce/Int64/dims=1L	86987 ns	87102 ns	1.00
array/reductions/reduce/Int64/dims=2L	84033 ns	84295 ns	1.00
array/reductions/reduce/Float32/1d	33717 ns	33785 ns	1.00
array/reductions/reduce/Float32/dims=1	39187 ns	48986 ns	0.80
array/reductions/reduce/Float32/dims=2	56332 ns	56655 ns	0.99
array/reductions/reduce/Float32/dims=1L	51325 ns	51438 ns	1.00
array/reductions/reduce/Float32/dims=2L	69258 ns	69460.5 ns	1.00
array/reductions/mapreduce/Int64/1d	39257.5 ns	38699 ns	1.01
array/reductions/mapreduce/Int64/dims=1	51465.5 ns	41686 ns	1.23
array/reductions/mapreduce/Int64/dims=2	58854 ns	58974 ns	1.00
array/reductions/mapreduce/Int64/dims=1L	87070 ns	87184 ns	1.00
array/reductions/mapreduce/Int64/dims=2L	84235 ns	84571 ns	1.00
array/reductions/mapreduce/Float32/1d	33106 ns	33512 ns	0.99
array/reductions/mapreduce/Float32/dims=1	39493.5 ns	47745 ns	0.83
array/reductions/mapreduce/Float32/dims=2	55782 ns	56241 ns	0.99
array/reductions/mapreduce/Float32/dims=1L	51188 ns	51435 ns	1.00
array/reductions/mapreduce/Float32/dims=2L	68300.5 ns	69604 ns	0.98
array/broadcast	20139 ns	20361 ns	0.99
array/copyto!/gpu_to_gpu	10570.666666666666 ns	10601.666666666666 ns	1.00
array/copyto!/cpu_to_gpu	213494 ns	214964 ns	0.99
array/copyto!/gpu_to_cpu	281531 ns	282717 ns	1.00
array/accumulate/Int64/1d	118212 ns	118054 ns	1.00
array/accumulate/Int64/dims=1	79130 ns	78929 ns	1.00
array/accumulate/Int64/dims=2	155192 ns	155861 ns	1.00
array/accumulate/Int64/dims=1L	1705618 ns	1705368 ns	1.00
array/accumulate/Int64/dims=2L	960326 ns	960330.5 ns	1.00
array/accumulate/Float32/1d	100301.5 ns	100426 ns	1.00
array/accumulate/Float32/dims=1	75752 ns	75943 ns	1.00
array/accumulate/Float32/dims=2	143887.5 ns	143974 ns	1.00
array/accumulate/Float32/dims=1L	1583879 ns	1584300 ns	1.00
array/accumulate/Float32/dims=2L	658657 ns	658063 ns	1.00
array/construct	1268.2 ns	1252.6 ns	1.01
array/random/randn/Float32	35400 ns	35435 ns	1.00
array/random/randn!/Float32	29917 ns	29972 ns	1.00
array/random/rand!/Int64	32291 ns	28260 ns	1.14
array/random/rand!/Float32	8209.333333333334 ns	8310 ns	0.99
array/random/rand/Int64	29169.5 ns	29927 ns	0.97
array/random/rand/Float32	12376.5 ns	12324 ns	1.00
array/permutedims/4d	51508 ns	51686 ns	1.00
array/permutedims/2d	52234.5 ns	52279 ns	1.00
array/permutedims/3d	52585 ns	52911 ns	0.99
array/sorting/1d	2733948 ns	2735042.5 ns	1.00
array/sorting/by	3303469 ns	3304486.5 ns	1.00
array/sorting/2d	1066285 ns	1066581 ns	1.00
cuda/synchronization/stream/auto	1039.7272727272727 ns	993.5882352941177 ns	1.05
cuda/synchronization/stream/nonblocking	7625.1 ns	7392.700000000001 ns	1.03
cuda/synchronization/stream/blocking	783.9074074074074 ns	811.8282828282828 ns	0.97
cuda/synchronization/context/auto	1145.4 ns	1160.9 ns	0.99
cuda/synchronization/context/nonblocking	7973.1 ns	7875.6 ns	1.01
cuda/synchronization/context/blocking	881.3770491803278 ns	899.7058823529412 ns	0.98

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

[maleadt]

I'll fold this into #3020.