Test loading of package on unsupported platforms

[christiangnrd]

Close #508

[codecov]

Codecov Report

Attention: Patch coverage is 57.14286% with 3 lines in your changes missing coverage. Please review.

Project coverage is 75.05%. Comparing base (52d7056) to head (a319488).
Report is 395 commits behind head on main.

Files with missing lines	Patch %	Lines
src/version.jl	66.66%	2 Missing ⚠️
src/state.jl	0.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #509      +/-   ##
==========================================
+ Coverage   71.04%   75.05%   +4.01%     
==========================================
  Files          36       57      +21     
  Lines        1143     2702    +1559     
==========================================
+ Hits          812     2028    +1216     
- Misses        331      674     +343     

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

[github-actions]

Metal Benchmarks

Benchmark suite	Current: a319488	Previous: 6ac7f3c	Ratio
private array/construct	26702.428571428572 ns	26131 ns	1.02
private array/broadcast	454916 ns	463417 ns	0.98
private array/random/randn/Float32	818833 ns	896416 ns	0.91
private array/random/randn!/Float32	641333 ns	591354.5 ns	1.08
private array/random/rand!/Int64	553292 ns	550958 ns	1.00
private array/random/rand!/Float32	592500 ns	550458 ns	1.08
private array/random/rand/Int64	776999.5 ns	892521 ns	0.87
private array/random/rand/Float32	571833.5 ns	776917 ns	0.74
private array/copyto!/gpu_to_gpu	694750 ns	594083.5 ns	1.17
private array/copyto!/cpu_to_gpu	802958 ns	658708 ns	1.22
private array/copyto!/gpu_to_cpu	630458 ns	649209 ns	0.97
private array/accumulate/1d	1318125 ns	1443834 ns	0.91
private array/accumulate/2d	1367354 ns	1500708 ns	0.91
private array/iteration/findall/int	2055833 ns	2321979 ns	0.89
private array/iteration/findall/bool	1813729 ns	2055750 ns	0.88
private array/iteration/findfirst/int	1674459 ns	1811916.5 ns	0.92
private array/iteration/findfirst/bool	1651021 ns	1734959 ns	0.95
private array/iteration/scalar	3525833 ns	2445625 ns	1.44
private array/iteration/logical	3145062.5 ns	3496146.5 ns	0.90
private array/iteration/findmin/1d	1754146 ns	1932458 ns	0.91
private array/iteration/findmin/2d	1334417 ns	1457562.5 ns	0.92
private array/reductions/reduce/1d	1016396.5 ns	970146 ns	1.05
private array/reductions/reduce/2d	649979 ns	729750 ns	0.89
private array/reductions/mapreduce/1d	1004875 ns	985166.5 ns	1.02
private array/reductions/mapreduce/2d	649792 ns	707166.5 ns	0.92
private array/permutedims/4d	2539917 ns	2696500 ns	0.94
private array/permutedims/2d	1007042 ns	1122916.5 ns	0.90
private array/permutedims/3d	1582666 ns	1861125 ns	0.85
private array/copy	593583 ns	809834 ns	0.73
latency/precompile	5744969750.5 ns	5830401479.5 ns	0.99
latency/ttfp	3042026375 ns	3083600896 ns	0.99
latency/import	1140101208 ns	1160891875 ns	0.98
integration/metaldevrt	704042 ns	760875 ns	0.93
integration/byval/slices=1	1530083 ns	1682042 ns	0.91
integration/byval/slices=3	9602916 ns	19850395.5 ns	0.48
integration/byval/reference	1555333.5 ns	1666916 ns	0.93
integration/byval/slices=2	2479250 ns	2824958 ns	0.88
kernel/indexing	470000 ns	460854.5 ns	1.02
kernel/indexing_checked	473250 ns	468479.5 ns	1.01
kernel/launch	36916.75 ns	8250 ns	4.47
metal/synchronization/stream	14333 ns	15042 ns	0.95
metal/synchronization/context	14417 ns	15375 ns	0.94
shared array/construct	26845.14285714286 ns	26416.666666666668 ns	1.02
shared array/broadcast	459520.5 ns	463291 ns	0.99
shared array/random/randn/Float32	817021 ns	933042 ns	0.88
shared array/random/randn!/Float32	665417 ns	589020.5 ns	1.13
shared array/random/rand!/Int64	558937.5 ns	551208 ns	1.01
shared array/random/rand!/Float32	603792 ns	559708 ns	1.08
shared array/random/rand/Int64	793166.5 ns	909896 ns	0.87
shared array/random/rand/Float32	605375 ns	823646 ns	0.73
shared array/copyto!/gpu_to_gpu	86292 ns	81542 ns	1.06
shared array/copyto!/cpu_to_gpu	88625 ns	82042 ns	1.08
shared array/copyto!/gpu_to_cpu	77000 ns	79958 ns	0.96
shared array/accumulate/1d	1323209 ns	1441770.5 ns	0.92
shared array/accumulate/2d	1369417 ns	1534062.5 ns	0.89
shared array/iteration/findall/int	1741250 ns	2050459 ns	0.85
shared array/iteration/findall/bool	1645854 ns	1778792 ns	0.93
shared array/iteration/findfirst/int	1373042 ns	1511104 ns	0.91
shared array/iteration/findfirst/bool	1347708 ns	1457771 ns	0.92
shared array/iteration/scalar	151084 ns	161334 ns	0.94
shared array/iteration/logical	2940292 ns	3295666.5 ns	0.89
shared array/iteration/findmin/1d	1463729.5 ns	1575500 ns	0.93
shared array/iteration/findmin/2d	1348458 ns	1465729 ns	0.92
shared array/reductions/reduce/1d	721354.5 ns	724292 ns	1.00
shared array/reductions/reduce/2d	659916 ns	704958 ns	0.94
shared array/reductions/mapreduce/1d	714250 ns	719478.5 ns	0.99
shared array/reductions/mapreduce/2d	657542 ns	710875 ns	0.92
shared array/permutedims/4d	2458084 ns	2675917 ns	0.92
shared array/permutedims/2d	1008729.5 ns	1127021 ns	0.90
shared array/permutedims/3d	1585250 ns	1878708.5 ns	0.84
shared array/copy	242958 ns	209292 ns	1.16

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

[maleadt]

Interesting approach. This should make it possible to remove the PkgEval blacklist, so that package loadability is also tested there. I wonder if we need to do something similar with CUDA.jl and the other back-ends.

[christiangnrd]

@maleadt I made the support check message more verbose for paravirtual devices.

I also added a helper function for safely checking the lower bound on all OSes as this'll become useful as we add more API wrappers for macos 14 and 15 features (like the MTLArchitecture wrapper I also snuck into this PR). Not duplicating all of MTLDevice to add the architecture::MTLArchitecture property that only exists since macOS 14 is what inspired JuliaInterop/ObjectiveC.jl#46

[christiangnrd]

Failure is the riscv bug

[christiangnrd]

Accidentally deleted branch, reopening.