Update buildkite for the 1.12 release

[kshyatt]

No description provided.

[github-actions]

CUDA.jl Benchmarks

Benchmark suite	Current: 4812987	Previous: f7deec6	Ratio
latency/precompile	56763592928.5 ns	56924777734.5 ns	1.00
latency/ttfp	8388724229 ns	8417873332.5 ns	1.00
latency/import	4509004428 ns	4531361015 ns	1.00
integration/volumerhs	9628927 ns	9625377 ns	1.00
integration/byval/slices=1	146968 ns	146827 ns	1.00
integration/byval/slices=3	426118 ns	425931 ns	1.00
integration/byval/reference	145043 ns	144949 ns	1.00
integration/byval/slices=2	286546 ns	286317 ns	1.00
integration/cudadevrt	103524.5 ns	103600 ns	1.00
kernel/indexing	14190 ns	14225 ns	1.00
kernel/indexing_checked	14880 ns	15087 ns	0.99
kernel/occupancy	666.8451612903226 ns	679.8954248366013 ns	0.98
kernel/launch	2188.8888888888887 ns	2150.3333333333335 ns	1.02
kernel/rand	15969 ns	14810 ns	1.08
array/reverse/1d	20201.5 ns	20182 ns	1.00
array/reverse/2dL_inplace	66796 ns	66832.5 ns	1.00
array/reverse/1dL	70434.5 ns	70358 ns	1.00
array/reverse/2d	22129 ns	21865 ns	1.01
array/reverse/1d_inplace	9629 ns	11480 ns	0.84
array/reverse/2d_inplace	13283 ns	13272 ns	1.00
array/reverse/2dL	74211 ns	73906 ns	1.00
array/reverse/1dL_inplace	66770.5 ns	66817 ns	1.00
array/copy	21070.5 ns	20949 ns	1.01
array/iteration/findall/int	158872 ns	157295 ns	1.01
array/iteration/findall/bool	140755 ns	139923.5 ns	1.01
array/iteration/findfirst/int	162420 ns	161193 ns	1.01
array/iteration/findfirst/bool	163256 ns	162272 ns	1.01
array/iteration/scalar	72491 ns	73738 ns	0.98
array/iteration/logical	217109 ns	214452.5 ns	1.01
array/iteration/findmin/1d	111541.5 ns	50889.5 ns	2.19
array/iteration/findmin/2d	109281.5 ns	96643 ns	1.13
array/reductions/reduce/Int64/1d	44330 ns	43989 ns	1.01
array/reductions/reduce/Int64/dims=1	45355.5 ns	44879 ns	1.01
array/reductions/reduce/Int64/dims=2	61954 ns	61825 ns	1.00
array/reductions/reduce/Int64/dims=1L	89305 ns	89232 ns	1.00
array/reductions/reduce/Int64/dims=2L	88461.5 ns	88384 ns	1.00
array/reductions/reduce/Float32/1d	37532 ns	37163 ns	1.01
array/reductions/reduce/Float32/dims=1	51358.5 ns	47666 ns	1.08
array/reductions/reduce/Float32/dims=2	60142 ns	59848 ns	1.00
array/reductions/reduce/Float32/dims=1L	52443 ns	52408 ns	1.00
array/reductions/reduce/Float32/dims=2L	72543 ns	72122.5 ns	1.01
array/reductions/mapreduce/Int64/1d	43945.5 ns	43666 ns	1.01
array/reductions/mapreduce/Int64/dims=1	44948 ns	47028 ns	0.96
array/reductions/mapreduce/Int64/dims=2	61902 ns	61661 ns	1.00
array/reductions/mapreduce/Int64/dims=1L	89150 ns	88863 ns	1.00
array/reductions/mapreduce/Int64/dims=2L	88467 ns	88192 ns	1.00
array/reductions/mapreduce/Float32/1d	37797 ns	37065 ns	1.02
array/reductions/mapreduce/Float32/dims=1	52170.5 ns	42446.5 ns	1.23
array/reductions/mapreduce/Float32/dims=2	60091 ns	60229 ns	1.00
array/reductions/mapreduce/Float32/dims=1L	52664 ns	52761 ns	1.00
array/reductions/mapreduce/Float32/dims=2L	72804 ns	72561 ns	1.00
array/broadcast	20148 ns	20011 ns	1.01
array/copyto!/gpu_to_gpu	11348 ns	11355.5 ns	1.00
array/copyto!/cpu_to_gpu	218691 ns	216192 ns	1.01
array/copyto!/gpu_to_cpu	283366 ns	283975.5 ns	1.00
array/accumulate/Int64/1d	125063 ns	125034.5 ns	1.00
array/accumulate/Int64/dims=1	83798 ns	83398 ns	1.00
array/accumulate/Int64/dims=2	158223 ns	157817 ns	1.00
array/accumulate/Int64/dims=1L	1710340 ns	1708490 ns	1.00
array/accumulate/Int64/dims=2L	967011 ns	966251 ns	1.00
array/accumulate/Float32/1d	109553 ns	109114 ns	1.00
array/accumulate/Float32/dims=1	81093.5 ns	80351 ns	1.01
array/accumulate/Float32/dims=2	147756.5 ns	147295.5 ns	1.00
array/accumulate/Float32/dims=1L	1619562.5 ns	1618020.5 ns	1.00
array/accumulate/Float32/dims=2L	698804 ns	698067 ns	1.00
array/construct	1268.8 ns	1296.7 ns	0.98
array/random/randn/Float32	46089 ns	48838.5 ns	0.94
array/random/randn!/Float32	25027 ns	24912 ns	1.00
array/random/rand!/Int64	27258 ns	27275 ns	1.00
array/random/rand!/Float32	8876.333333333334 ns	8805.333333333334 ns	1.01
array/random/rand/Int64	29951.5 ns	30044 ns	1.00
array/random/rand/Float32	13476.5 ns	13354 ns	1.01
array/permutedims/4d	59981 ns	60446 ns	0.99
array/permutedims/2d	54243 ns	54105.5 ns	1.00
array/permutedims/3d	54968 ns	54893 ns	1.00
array/sorting/1d	2758013 ns	2756483.5 ns	1.00
array/sorting/by	3347816 ns	3368977 ns	0.99
array/sorting/2d	1088878 ns	1088064.5 ns	1.00
cuda/synchronization/stream/auto	1008.5833333333334 ns	1030.1 ns	0.98
cuda/synchronization/stream/nonblocking	7512.700000000001 ns	7504.6 ns	1.00
cuda/synchronization/stream/blocking	825.3168316831683 ns	801.2842105263157 ns	1.03
cuda/synchronization/context/auto	1154.6 ns	1179.3 ns	0.98
cuda/synchronization/context/nonblocking	7388.4 ns	7293.5 ns	1.01
cuda/synchronization/context/blocking	913.765306122449 ns	909.9636363636364 ns	1.00

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

[maleadt]

[61eb1bfa] GPUCompiler v1.1.0

Why did this install an ancient version of GPUCompiler...

[kshyatt]

No idea. But now that this is an official release we should anyway use Resolver. Let me add that.

[kshyatt]

Oh wait, but with nightly gone from the version tag, it is using Resolver...

[kshyatt]

Oh it's doing it because now it is using Resolver and our compat entry for GPUCompiler says GPUCompiler = "1.1" lol

[maleadt]

I'll check which version of GPUCompiler.jl works on 1.12 and edit the compat bounds in General.
EDIT: JuliaRegistries/General#140014

[maleadt]

Let's try increasing the bound on GPUCompiler.jl (e.g. from v1.6.0 as imposed by General to v1.7?) to see if this is a bug that's been fixed since.

[kshyatt]

Probably did this wrong but it's still using 1.6...

[maleadt]

Ugh, so this looks like a legitimate bug.

[maleadt]

I can't reproduce the crash easily, but looking into it I came across a hang (JuliaLang/julia#59834), so first looking at that.