use device libs from AMDGPU_LLVM_Backend_jll (2)

[simeonschaub]

We downgrade these into a scratch space to be compatible with Julia's LLVM version

ref #919 (comment)

Reopened to run MI300 CI

[github-actions]

AMDGPU.jl Benchmarks

Details

Benchmark suite	Current: e093932	Previous: 58dd394	Ratio
amdgpu/synchronization/context/device	637.5 ns	540 ns	1.18
amdgpu/synchronization/stream/blocking	262.5 ns	227.5 ns	1.15
amdgpu/synchronization/stream/nonblocking	370 ns	310 ns	1.19
array/accumulate/Float32/1d	73528.5 ns	78001 ns	0.94
array/accumulate/Float32/dims=1	279739 ns	275066.5 ns	1.02
array/accumulate/Float32/dims=1L	81218.5 ns	82516.25 ns	0.98
array/accumulate/Float32/dims=2	80671 ns	81193.75 ns	0.99
array/accumulate/Float32/dims=2L	2764466.25 ns	2758054.75 ns	1.00
array/accumulate/Int64/1d	79143.5 ns	79148.5 ns	1.00
array/accumulate/Int64/dims=1	246286 ns	243553.5 ns	1.01
array/accumulate/Int64/dims=1L	84186.25 ns	84101.25 ns	1.00
array/accumulate/Int64/dims=2	86363.75 ns	86991.25 ns	0.99
array/accumulate/Int64/dims=2L	2893263 ns	2895916.75 ns	1.00
array/broadcast	73411 ns	74006 ns	0.99
array/construct	2327.5 ns	2335 ns	1.00
array/copy	36795.5 ns	37160.75 ns	0.99
array/copyto!/cpu_to_gpu	111546.5 ns	111339.25 ns	1.00
array/copyto!/gpu_to_cpu	111924 ns	111596.5 ns	1.00
array/copyto!/gpu_to_gpu	48595.75 ns	59505.75 ns	0.82
array/iteration/findall/bool	140069.5 ns	139364.75 ns	1.01
array/iteration/findall/int	151072.25 ns	152744.75 ns	0.99
array/iteration/findfirst/bool	146529.75 ns	145527.25 ns	1.01
array/iteration/findfirst/int	145699.5 ns	145997.25 ns	1.00
array/iteration/findmin/1d	121736.75 ns	124581.75 ns	0.98
array/iteration/findmin/2d	110106.5 ns	98116.25 ns	1.12
array/iteration/logical	252573.5 ns	252929 ns	1.00
array/iteration/scalar	300054.25 ns	309789.75 ns	0.97
array/permutedims/2d	71786 ns	72443.5 ns	0.99
array/permutedims/3d	71283.5 ns	71971 ns	0.99
array/permutedims/4d	72348.5 ns	75061 ns	0.96
array/random/rand/Float32	45133.25 ns	45963 ns	0.98
array/random/rand/Int64	54388.25 ns	54770.75 ns	0.99
array/random/rand!/Float32	65518.5 ns	44300.75 ns	1.48
array/random/rand!/Int64	73301 ns	73566 ns	1.00
array/random/randn/Float32	73111 ns	76658.75 ns	0.95
array/random/randn!/Float32	80838.5 ns	81353.5 ns	0.99
array/reductions/mapreduce/Float32/1d	100979 ns	101831.5 ns	0.99
array/reductions/mapreduce/Float32/dims=1	89686.25 ns	91791.25 ns	0.98
array/reductions/mapreduce/Float32/dims=1L	832701.75 ns	834629.75 ns	1.00
array/reductions/mapreduce/Float32/dims=2	87598.75 ns	88776.5 ns	0.99
array/reductions/mapreduce/Float32/dims=2L	144267 ns	144237 ns	1.00
array/reductions/mapreduce/Int64/1d	100358.75 ns	101501.5 ns	0.99
array/reductions/mapreduce/Int64/dims=1	89498.75 ns	90426.25 ns	0.99
array/reductions/mapreduce/Int64/dims=1L	830816.75 ns	834257.25 ns	1.00
array/reductions/mapreduce/Int64/dims=2	86561.25 ns	88551.5 ns	0.98
array/reductions/mapreduce/Int64/dims=2L	145214.5 ns	145772.25 ns	1.00
array/reductions/reduce/Float32/1d	99724 ns	102909 ns	0.97
array/reductions/reduce/Float32/dims=1	89713.75 ns	91134 ns	0.98
array/reductions/reduce/Float32/dims=1L	832726.75 ns	835889.75 ns	1.00
array/reductions/reduce/Float32/dims=2	87591.25 ns	88548.75 ns	0.99
array/reductions/reduce/Float32/dims=2L	144577.25 ns	145259.5 ns	1.00
array/reductions/reduce/Int64/1d	100024 ns	101701.5 ns	0.98
array/reductions/reduce/Int64/dims=1	88136.25 ns	88971.5 ns	0.99
array/reductions/reduce/Int64/dims=1L	831411.75 ns	831467.25 ns	1.00
array/reductions/reduce/Int64/dims=2	87443.75 ns	88508.75 ns	0.99
array/reductions/reduce/Int64/dims=2L	144917.25 ns	144032.25 ns	1.01
array/reverse/1d	45500.5 ns	46513.25 ns	0.98
array/reverse/1dL	73598.5 ns	73918.5 ns	1.00
array/reverse/1dL_inplace	79738.5 ns	81293.75 ns	0.98
array/reverse/1d_inplace	52068.25 ns	53618.5 ns	0.97
array/reverse/2d	50195.75 ns	51231 ns	0.98
array/reverse/2dL	84441.25 ns	84231.25 ns	1.00
array/reverse/2dL_inplace	69383.5 ns	92779 ns	0.75
array/reverse/2d_inplace	63363.25 ns	42753 ns	1.48
array/sorting/1d	333994.5 ns	334009.75 ns	1.00
integration/byval/reference	39871 ns	39821 ns	1.00
integration/byval/slices=1	40820 ns	40321 ns	1.01
integration/byval/slices=2	140692 ns	147452 ns	0.95
integration/byval/slices=3	237083 ns	240194 ns	0.99
integration/volumerhs	5011010 ns	5007121 ns	1.00
kernel/indexing	48793.25 ns	58288.5 ns	0.84
kernel/indexing_checked	48423.25 ns	50148.25 ns	0.97
kernel/launch	1480 ns	1445 ns	1.02
kernel/rand	100961.5 ns	101319 ns	1.00
latency/import	1693628312 ns	1689718608 ns	1.00
latency/precompile	39192093618 ns	38551162270 ns	1.02
latency/ttfp	2314342244 ns	2308017436 ns	1.00

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

[simeonschaub]

OK, I think this should be good to go now