Ignore NVIDIA driver store CUDA libraries from system path warnings on Windows

[huiyuxie]

Fix #2836. I‘ve also run into this issue when I update my NVIDIA driver to the latest release. And this warning is so annoying so I prefer to fix it myself.

julia> using CUDA
┌ Warning: CUDA runtime library `nvcudart_hybrid64.dll` was loaded from a system path, `C:\WINDOWS\System32\DriverStore\FileRepository\nvdmi.inf_amd64_84c88fdcdcf1f878\nvcudart_hybrid64.dll`.
│ This may cause errors.
│
│ If you're running under a profiler, this situation is expected. Otherwise,
│ ensure that your library path environment variable (e.g., `PATH` on Windows
│ or `LD_LIBRARY_PATH` on Linux) does not include CUDA library paths.
│
│ In any other case, please file an issue.
└ @ CUDA C:\Users\huiyu\.julia\packages\CUDA\Wfi8S\src\initialization.jl:218

Similar warning from my CUDA.jl but not exactly the same path.

Review required @maleadt.

[github-actions]

Your PR requires formatting changes to meet the project's style guidelines.
Please consider running Runic (git runic master) to apply these changes.

Click here to view the suggested changes.

diff --git a/src/initialization.jl b/src/initialization.jl
index 0574527b8..41a6aaa86 100644
--- a/src/initialization.jl
+++ b/src/initialization.jl
@@ -214,13 +214,13 @@ function __init__()
                              "cublas", "cupti", "cusparse", "cufft", "curand", "cusolver"]
         for lib in Libdl.dllist()
             contains(lib, "artifacts") && continue
-            
+
             # skip driver store directories on Windows - these contain legitimate libraries
             # that are part of the display driver installation (at least on CUDA 13+)
             if Sys.iswindows() && contains(lib, "DriverStore")
                 continue
             end
-            
+
             if any(rtlib -> contains(lib, rtlib), runtime_libraries)
                 @warn """CUDA runtime library `$(basename(lib))` was loaded from a system path, `$lib`.
                          This may cause errors.

[codecov]

Codecov Report

❌ Patch coverage is 50.00000% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 89.79%. Comparing base (d6ad9c3) to head (e006c4d).

Files with missing lines	Patch %	Lines
src/initialization.jl	50.00%	1 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           master    #2847       +/-   ##
===========================================
+ Coverage   77.39%   89.79%   +12.39%     
===========================================
  Files         150      150               
  Lines       13124    13235      +111     
===========================================
+ Hits        10157    11884     +1727     
+ Misses       2967     1351     -1616     

☔ 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.

[kshyatt]

@huiyuxie Tim will be away for a bit. Maybe @vchuravy could review?

[huiyuxie]

Thanks for letting me know!

Maybe @vchuravy could review?

That will be good.

[github-actions]

CUDA.jl Benchmarks

Benchmark suite	Current: be4bfbe	Previous: e734d7a	Ratio
latency/precompile	43525633047 ns	43398695331.5 ns	1.00
latency/ttfp	7185459930 ns	7160470719 ns	1.00
latency/import	3669446977 ns	3655684127 ns	1.00
integration/volumerhs	9607576 ns	9617572.5 ns	1.00
integration/byval/slices=1	147348 ns	147149 ns	1.00
integration/byval/slices=3	426210 ns	426071 ns	1.00
integration/byval/reference	145242 ns	145099 ns	1.00
integration/byval/slices=2	286637 ns	286606 ns	1.00
integration/cudadevrt	103671 ns	103567 ns	1.00
kernel/indexing	14499.5 ns	14309 ns	1.01
kernel/indexing_checked	15224 ns	14963.5 ns	1.02
kernel/occupancy	688.2733333333333 ns	667.6981132075472 ns	1.03
kernel/launch	2191 ns	2240.222222222222 ns	0.98
kernel/rand	17993 ns	15887 ns	1.13
array/reverse/1d	20658 ns	19850 ns	1.04
array/reverse/2d	24952 ns	25206 ns	0.99
array/reverse/1d_inplace	11049 ns	10666 ns	1.04
array/reverse/2d_inplace	12495 ns	12171 ns	1.03
array/copy	21475.5 ns	20937 ns	1.03
array/iteration/findall/int	159129 ns	157262 ns	1.01
array/iteration/findall/bool	140150.5 ns	139199 ns	1.01
array/iteration/findfirst/int	160347 ns	2147909.5 ns	0.07465258661968766
array/iteration/findfirst/bool	161398 ns	2129592 ns	0.07578822610152555
array/iteration/scalar	74110 ns	71676 ns	1.03
array/iteration/logical	218224.5 ns	235260.5 ns	0.93
array/iteration/findmin/1d	47450 ns	239425 ns	0.20
array/iteration/findmin/2d	97634 ns	96120 ns	1.02
array/reductions/reduce/Int64/1d	43795.5 ns	147373 ns	0.30
array/reductions/reduce/Int64/dims=1	50161 ns	44089 ns	1.14
array/reductions/reduce/Int64/dims=2	62912 ns	61492 ns	1.02
array/reductions/reduce/Int64/dims=1L	89385 ns	89053 ns	1.00
array/reductions/reduce/Int64/dims=2L	89067.5 ns	661490 ns	0.13
array/reductions/reduce/Float32/1d	36095 ns	104099.5 ns	0.35
array/reductions/reduce/Float32/dims=1	52477 ns	40848 ns	1.28
array/reductions/reduce/Float32/dims=2	60415 ns	59344 ns	1.02
array/reductions/reduce/Float32/dims=1L	52613 ns	52528 ns	1.00
array/reductions/reduce/Float32/dims=2L	71055 ns	547809 ns	0.13
array/reductions/mapreduce/Int64/1d	44210.5 ns	149021 ns	0.30
array/reductions/mapreduce/Int64/dims=1	47420 ns	44144 ns	1.07
array/reductions/mapreduce/Int64/dims=2	62922 ns	61760 ns	1.02
array/reductions/mapreduce/Int64/dims=1L	89320 ns	89012 ns	1.00
array/reductions/mapreduce/Int64/dims=2L	88279 ns	685323 ns	0.13
array/reductions/mapreduce/Float32/1d	36180 ns	104623 ns	0.35
array/reductions/mapreduce/Float32/dims=1	42363 ns	41094 ns	1.03
array/reductions/mapreduce/Float32/dims=2	60733 ns	59797 ns	1.02
array/reductions/mapreduce/Float32/dims=1L	53138 ns	52788.5 ns	1.01
array/reductions/mapreduce/Float32/dims=2L	71222 ns	549999.5 ns	0.13
array/broadcast	20384.5 ns	20109 ns	1.01
array/copyto!/gpu_to_gpu	11483 ns	12888 ns	0.89
array/copyto!/cpu_to_gpu	217325 ns	213909 ns	1.02
array/copyto!/gpu_to_cpu	286542 ns	283768.5 ns	1.01
array/accumulate/Int64/1d	125772.5 ns	124986 ns	1.01
array/accumulate/Int64/dims=1	84668 ns	83520 ns	1.01
array/accumulate/Int64/dims=2	159906 ns	158891 ns	1.01
array/accumulate/Int64/dims=1L	1721564 ns	1720572.5 ns	1.00
array/accumulate/Int64/dims=2L	968748.5 ns	967934 ns	1.00
array/accumulate/Float32/1d	110138 ns	109201 ns	1.01
array/accumulate/Float32/dims=1	81350 ns	80412 ns	1.01
array/accumulate/Float32/dims=2	148437 ns	147560.5 ns	1.01
array/accumulate/Float32/dims=1L	1619419 ns	1618433 ns	1.00
array/accumulate/Float32/dims=2L	698806 ns	698180 ns	1.00
array/construct	1279.55 ns	1627.3 ns	0.79
array/random/randn/Float32	47994 ns	45289.5 ns	1.06
array/random/randn!/Float32	25509 ns	25002 ns	1.02
array/random/rand!/Int64	27585 ns	27438 ns	1.01
array/random/rand!/Float32	8971.333333333334 ns	8965 ns	1.00
array/random/rand/Int64	37903.5 ns	29907 ns	1.27
array/random/rand/Float32	13636 ns	13209 ns	1.03
array/permutedims/4d	60796 ns	60267 ns	1.01
array/permutedims/2d	54865.5 ns	53766.5 ns	1.02
array/permutedims/3d	55580 ns	54703 ns	1.02
array/sorting/1d	2758924 ns	2755967.5 ns	1.00
array/sorting/by	3345169 ns	3342542 ns	1.00
array/sorting/2d	1081246 ns	1079574 ns	1.00
cuda/synchronization/stream/auto	1065.7 ns	1015.3 ns	1.05
cuda/synchronization/stream/nonblocking	7771.8 ns	7605.6 ns	1.02
cuda/synchronization/stream/blocking	857.3880597014926 ns	809.5368421052632 ns	1.06
cuda/synchronization/context/auto	1212.8 ns	1144.9 ns	1.06
cuda/synchronization/context/nonblocking	7095.9 ns	8738.6 ns	0.81
cuda/synchronization/context/blocking	922.9024390243902 ns	884.6666666666666 ns	1.04

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