Adapt to LLVMDowngrader_jll 0.11 and GPUCompiler 2.8

[maleadt]

Bumps the compat bounds to LLVMDowngrader_jll 0.11 (JuliaPackaging/Yggdrasil#14758, built against LLVM 23) and GPUCompiler 2.8 (JuliaGPU/GPUCompiler.jl#931). The downgrader C API is unchanged, so this is compat-only.

[github-actions]

Metal Benchmarks

Details

Benchmark suite	Current: f498a72	Previous: 8b6fa21	Ratio
array/accumulate/Float32/1d	405083 ns	408042 ns	0.99
array/accumulate/Float32/dims=1	378458 ns	375167 ns	1.01
array/accumulate/Float32/dims=1L	8802625 ns	8807041 ns	1.00
array/accumulate/Float32/dims=2	441167 ns	441625 ns	1.00
array/accumulate/Float32/dims=2L	2595750 ns	2593291 ns	1.00
array/accumulate/Int64/1d	839875 ns	842500 ns	1.00
array/accumulate/Int64/dims=1	937709 ns	708375 ns	1.32
array/accumulate/Int64/dims=1L	9517542 ns	9500083 ns	1.00
array/accumulate/Int64/dims=2	1240708 ns	1243500 ns	1.00
array/accumulate/Int64/dims=2L	6531708 ns	6527833 ns	1.00
array/broadcast	223167 ns	231250 ns	0.97
array/construct	2250 ns	2208 ns	1.02
array/permutedims/2d	400375 ns	398708 ns	1.00
array/permutedims/3d	1018750 ns	1015166 ns	1.00
array/permutedims/4d	1154125 ns	1206375 ns	0.96
array/private/copy	232208 ns	235708 ns	0.99
array/private/copyto!/cpu_to_gpu	198875 ns	199375 ns	1.00
array/private/copyto!/gpu_to_cpu	199625 ns	198541 ns	1.01
array/private/copyto!/gpu_to_gpu	196625 ns	198333 ns	0.99
array/private/iteration/findall/bool	1063583 ns	1067167 ns	1.00
array/private/iteration/findall/int	1222042 ns	1224083 ns	1.00
array/private/iteration/findfirst/bool	1095750 ns	910833 ns	1.20
array/private/iteration/findfirst/int	1137708 ns	1136500 ns	1.00
array/private/iteration/findmin/1d	1210167 ns	1221417 ns	0.99
array/private/iteration/findmin/2d	1053083 ns	1048917 ns	1.00
array/private/iteration/logical	1710416 ns	1708083 ns	1.00
array/private/iteration/scalar	1228500 ns	1286125 ns	0.96
array/random/rand/Float32	426042 ns	430833 ns	0.99
array/random/rand/Int64	518125 ns	520666 ns	1.00
array/random/rand!/Float32	380625 ns	384792 ns	0.99
array/random/rand!/Int64	422959 ns	404125 ns	1.05
array/random/randn/Float32	400166 ns	402459 ns	0.99
array/random/randn!/Float32	351167 ns	292542 ns	1.20
array/reductions/mapreduce/Float32/1d	408291 ns	424625 ns	0.96
array/reductions/mapreduce/Float32/dims=1	338625 ns	343417 ns	0.99
array/reductions/mapreduce/Float32/dims=1L	624625 ns	629792 ns	0.99
array/reductions/mapreduce/Float32/dims=2	348916 ns	351250 ns	0.99
array/reductions/mapreduce/Float32/dims=2L	1073542 ns	1062333 ns	1.01
array/reductions/mapreduce/Int64/1d	543584 ns	608667 ns	0.89
array/reductions/mapreduce/Int64/dims=1	629958 ns	454208 ns	1.39
array/reductions/mapreduce/Int64/dims=1L	1039583 ns	1027542 ns	1.01
array/reductions/mapreduce/Int64/dims=2	784833 ns	785375 ns	1.00
array/reductions/mapreduce/Int64/dims=2L	2189542 ns	2193208 ns	1.00
array/reductions/reduce/Float32/1d	433375 ns	441208 ns	0.98
array/reductions/reduce/Float32/dims=1	289083 ns	290917 ns	0.99
array/reductions/reduce/Float32/dims=1L	629916 ns	631750 ns	1.00
array/reductions/reduce/Float32/dims=2	233084 ns	229667 ns	1.01
array/reductions/reduce/Float32/dims=2L	452959 ns	450875 ns	1.00
array/reductions/reduce/Int64/1d	604084 ns	611417 ns	0.99
array/reductions/reduce/Int64/dims=1	459875 ns	636542 ns	0.72
array/reductions/reduce/Int64/dims=1L	1032167 ns	1031291 ns	1.00
array/reductions/reduce/Int64/dims=2	243875 ns	238084 ns	1.02
array/reductions/reduce/Int64/dims=2L	644417 ns	646084 ns	1.00
array/shared/copy	136208 ns	136292 ns	1.00
array/shared/copyto!/cpu_to_gpu	37916 ns	38125 ns	0.99
array/shared/copyto!/gpu_to_cpu	37625 ns	37625 ns	1
array/shared/copyto!/gpu_to_gpu	38000 ns	38208 ns	0.99
array/shared/iteration/findall/bool	1068000 ns	1097000 ns	0.97
array/shared/iteration/findall/int	1215167 ns	1232125 ns	0.99
array/shared/iteration/findfirst/bool	648958 ns	959417 ns	0.68
array/shared/iteration/findfirst/int	960375 ns	675750 ns	1.42
array/shared/iteration/findmin/1d	1081917 ns	1078750 ns	1.00
array/shared/iteration/findmin/2d	1053500 ns	1053000 ns	1.00
array/shared/iteration/logical	1564625 ns	1549833 ns	1.01
array/shared/iteration/scalar	3744.875 ns	3765.625 ns	0.99
array/sorting/1d	2089083 ns	2117166 ns	0.99
array/sorting/2d	8308584 ns	8292708 ns	1.00
integration/byval/reference	1104125 ns	1103833 ns	1.00
integration/byval/slices=1	1113042 ns	1106541 ns	1.01
integration/byval/slices=2	2020708 ns	2017500 ns	1.00
integration/byval/slices=3	6507459 ns	6579333 ns	0.99
integration/metaldevrt	381209 ns	378417 ns	1.01
kernel/indexing	184458 ns	210375 ns	0.88
kernel/indexing_checked	351875 ns	372375 ns	0.94
kernel/launch	1858.3 ns	1895.8 ns	0.98
kernel/rand	313208 ns	373333 ns	0.84
latency/import	1797781916 ns	1791487000 ns	1.00
latency/precompile	26253956291 ns	26158613125 ns	1.00
latency/ttfp	2300732333 ns	2293400750 ns	1.00
metal/synchronization/context	561.1559139784946 ns	556 ns	1.01
metal/synchronization/stream	354.07042253521126 ns	345.9859813084112 ns	1.02

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

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 86.71%. Comparing base (094a2d3) to head (f498a72).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #965      +/-   ##
==========================================
+ Coverage   86.69%   86.71%   +0.01%     
==========================================
  Files          77       77              
  Lines        5488     5488              
==========================================
+ Hits         4758     4759       +1     
+ Misses        730      729       -1     

☔ View full report in Codecov by Harness.
📢 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.