Skip to content

CUDA: Select compiler artifacts independently - #3227

Merged
maleadt merged 4 commits into
mainfrom
tb/compiler
Aug 10, 2026
Merged

CUDA: Select compiler artifacts independently#3227
maleadt merged 4 commits into
mainfrom
tb/compiler

Conversation

@maleadt

@maleadt maleadt commented Aug 10, 2026

Copy link
Copy Markdown
Member

Require CUDA_Compiler_jll 0.5 and CUDA_Runtime_jll 0.24, configure both version/local preference namespaces together, and route local compiler use through CUDA_Runtime_Discovery.

Add CUDA 13.4 to CI and exercise the newest compiler within each runtime major: 12.9 for CUDA 12 and the 13.4 EA compiler for CUDA 13. Verify that CI loads the requested compiler version.

maleadt added 4 commits August 9, 2026 17:38
@maleadt
CUDA: Select compiler artifacts independently
67e0113 
Require CUDA_Compiler_jll 0.5 and CUDA_Runtime_jll 0.24, configure both version/local preference namespaces together, and route local compiler use through CUDA_Runtime_Discovery.

Add CUDA 13.4 to CI and exercise the newest compiler within each runtime major: 12.9 for CUDA 12 and the 13.4 EA compiler for CUDA 13. Verify that CI loads the requested compiler version.
@maleadt
Remove version test.
a09358c
@maleadt
Document preference.
d095193
@maleadt
Fix pipeline.
48d902a 
[skip julia]
@codecov

codecov Bot commented Aug 10, 2026

Copy link
Copy Markdown

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.73%. Comparing base (959f3ca) to head (48d902a).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files 
@@            Coverage Diff             @@
##             main    #3227      +/-   ##
==========================================
- Coverage   92.25%   90.73%   -1.53%     
==========================================
  Files         184      184              
  Lines       15903    15901       -2     
==========================================
- Hits        14672    14428     -244     
- Misses       1231     1473     +242

☔ 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.

@github-actions

github-actions Bot commented Aug 10, 2026

Copy link
Copy Markdown
Contributor

CUDA.jl Benchmarks

Details
Benchmark suite Current: 48d902a Previous: afa5577 Ratio
array/accumulate/Float32/1d 98190 ns 97975 ns 1.00
array/accumulate/Float32/dims=1 71385 ns 71913 ns 0.99
array/accumulate/Float32/dims=1L 1598667 ns 1599238 ns 1.00
array/accumulate/Float32/dims=2 136565 ns 137770 ns 0.99
array/accumulate/Float32/dims=2L 659814 ns 660981 ns 1.00
array/accumulate/Int64/1d 117228 ns 117562 ns 1.00
array/accumulate/Int64/dims=1 75330 ns 75387 ns 1.00
array/accumulate/Int64/dims=1L 1715899 ns 1714776 ns 1.00
array/accumulate/Int64/dims=2 147837 ns 149123 ns 0.99
array/accumulate/Int64/dims=2L 986093 ns 985886 ns 1.00
array/broadcast 17609 ns 17732 ns 0.99
array/broadcast launch 8317 ns 8401.666666666666 ns 0.99
array/construct 835.3333333333334 ns 886.3695652173913 ns 0.94
array/copy 16347 ns 16274 ns 1.00
array/copyto!/cpu_to_gpu 209992 ns 206302 ns 1.02
array/copyto!/gpu_to_cpu 240386 ns 239976 ns 1.00
array/copyto!/gpu_to_gpu 8858 ns 8710 ns 1.02
array/iteration/findall/bool 130740 ns 130745 ns 1.00
array/iteration/findall/int 143888 ns 144920 ns 0.99
array/iteration/findfirst/bool 68240 ns 68050 ns 1.00
array/iteration/findfirst/int 69797 ns 69024 ns 1.01
array/iteration/findmin/1d 61564 ns 67713 ns 0.91
array/iteration/findmin/2d 99080 ns 98339 ns 1.01
array/iteration/logical 182287 ns 182409 ns 1.00
array/iteration/scalar 62721 ns 59585 ns 1.05
array/permutedims/2d 47436 ns 47775 ns 0.99
array/permutedims/3d 48972 ns 48890 ns 1.00
array/permutedims/4d 48922 ns 49194 ns 0.99
array/random/rand/Float32 11565 ns 11801 ns 0.98
array/random/rand/Int64 22664 ns 22532 ns 1.01
array/random/rand!/Float32 7836 ns 7627 ns 1.03
array/random/rand!/Int64 19783 ns 19605 ns 1.01
array/random/randn/Float32 32479 ns 32374 ns 1.00
array/random/randn!/Float32 23373 ns 23165 ns 1.01
array/reductions/mapreduce/Float32/1d 31688 ns 31664 ns 1.00
array/reductions/mapreduce/Float32/dims=1 37155 ns 37258 ns 1.00
array/reductions/mapreduce/Float32/dims=1L 49899 ns 50124 ns 1.00
array/reductions/mapreduce/Float32/dims=2 54566 ns 54553 ns 1.00
array/reductions/mapreduce/Float32/dims=2L 66421 ns 66620 ns 1.00
array/reductions/mapreduce/Int64/1d 38233 ns 37884 ns 1.01
array/reductions/mapreduce/Int64/dims=1 39846 ns 40141 ns 0.99
array/reductions/mapreduce/Int64/dims=1L 87893 ns 87814 ns 1.00
array/reductions/mapreduce/Int64/dims=2 57084 ns 57040 ns 1.00
array/reductions/mapreduce/Int64/dims=2L 82616 ns 82473 ns 1.00
array/reductions/reduce/Float32/1d 31686 ns 31361 ns 1.01
array/reductions/reduce/Float32/dims=1 37206 ns 37381 ns 1.00
array/reductions/reduce/Float32/dims=1L 49790 ns 50057 ns 0.99
array/reductions/reduce/Float32/dims=2 54509 ns 54746 ns 1.00
array/reductions/reduce/Float32/dims=2L 68029 ns 68086 ns 1.00
array/reductions/reduce/Int64/1d 38783 ns 38487 ns 1.01
array/reductions/reduce/Int64/dims=1 40140 ns 39867 ns 1.01
array/reductions/reduce/Int64/dims=1L 87798 ns 87918 ns 1.00
array/reductions/reduce/Int64/dims=2 56707 ns 56979 ns 1.00
array/reductions/reduce/Int64/dims=2L 82551 ns 82475 ns 1.00
array/reverse/1d 16332 ns 16193 ns 1.01
array/reverse/1dL 69045 ns 69089 ns 1.00
array/reverse/1dL_inplace 67011 ns 67143 ns 1.00
array/reverse/1d_inplace 8144 ns 8146.666666666667 ns 1.00
array/reverse/2d 19490 ns 19783 ns 0.99
array/reverse/2dL 72761 ns 73052 ns 1.00
array/reverse/2dL_inplace 66795 ns 66843 ns 1.00
array/reverse/2d_inplace 9421 ns 9531 ns 0.99
array/sorting/1d 2639397 ns 2637712 ns 1.00
array/sorting/2d 1027589 ns 1027828 ns 1.00
array/sorting/by 3182160 ns 3180438 ns 1.00
cuda/synchronization/context/auto 1014.8 ns 1005.9 ns 1.01
cuda/synchronization/context/blocking 782.8686868686868 ns 781.5588235294117 ns 1.00
cuda/synchronization/context/nonblocking 5630.833333333333 ns 5678 ns 0.99
cuda/synchronization/stream/auto 857.125 ns 861.3050847457627 ns 1.00
cuda/synchronization/stream/blocking 665.5194805194806 ns 664.2784810126582 ns 1.00
cuda/synchronization/stream/nonblocking 5543.285714285715 ns 5491.857142857143 ns 1.01
integration/byval/reference 147306 ns 147351 ns 1.00
integration/byval/slices=1 151283 ns 151176 ns 1.00
integration/byval/slices=2 294468 ns 294440 ns 1.00
integration/byval/slices=3 437479 ns 437439 ns 1.00
integration/cudadevrt 104462 ns 104621 ns 1.00
integration/volumerhs 9147008 ns 9143640 ns 1.00
kernel/indexing 12643 ns 12577 ns 1.01
kernel/indexing_checked 13323 ns 13304 ns 1.00
kernel/launch 1986.6666666666667 ns 2106.3333333333335 ns 0.94
kernel/occupancy 671.6149068322982 ns 668.2767295597484 ns 1.00
kernel/rand 13597 ns 13986 ns 0.97
latency/import 4381929737 ns 4375777011 ns 1.00
latency/precompile 4996073786 ns 5006135691 ns 1.00
latency/ttfp 5412596584 ns 5435893153 ns 1.00

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

@maleadt
maleadt merged commit 9194b55 into main Aug 10, 2026
1 check passed
@maleadt
maleadt deleted the tb/compiler branch August 10, 2026 21:07
@christiangnrd christiangnrd mentioned this pull request Aug 21, 2026
Merged
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

1 participant

@maleadt