Setup AMD MI300 CI runners on Alps

[luraess]

This PR enables AMDGPU CI on CSCS's ALPS MI300 partition.

A few things to note:

• It runs the standard test suite on Julia 1.12 and skips Enzyme tests for now.
• It runs the multi-device tests (2 GPUs)
• It runs on resources from the ∂GPU4GEO c44 project.
• It may not be triggered automatically from forks.

I suggest to merge this and see how things evolve over the coming 2-3 years after what we can decide how to continue.

[github-actions]

AMDGPU.jl Benchmarks

Details

Benchmark suite	Current: 4c35fe2	Previous: f49923a	Ratio
amdgpu/synchronization/context/device	610 ns	620 ns	0.98
amdgpu/synchronization/stream/blocking	250 ns	250 ns	1
amdgpu/synchronization/stream/nonblocking	340 ns	340 ns	1
array/accumulate/Float32/1d	86291 ns	86721 ns	1.00
array/accumulate/Float32/dims=1	399506 ns	283534 ns	1.41
array/accumulate/Float32/dims=1L	134782 ns	132701 ns	1.02
array/accumulate/Float32/dims=2	127522 ns	128622 ns	0.99
array/accumulate/Float32/dims=2L	2829850 ns	2830679 ns	1.00
array/accumulate/Int64/1d	97362 ns	96522 ns	1.01
array/accumulate/Int64/dims=1	283554 ns	398946 ns	0.71
array/accumulate/Int64/dims=1L	167543 ns	161462 ns	1.04
array/accumulate/Int64/dims=2	122532 ns	125241 ns	0.98
array/accumulate/Int64/dims=2L	3011602 ns	3010532 ns	1.00
array/broadcast	99432 ns	87271 ns	1.14
array/construct	1771 ns	1610 ns	1.10
array/copy	39741 ns	39990 ns	0.99
array/copyto!/cpu_to_gpu	183012 ns	183233 ns	1.00
array/copyto!/gpu_to_cpu	114942 ns	119621 ns	0.96
array/copyto!/gpu_to_gpu	66981 ns	66311 ns	1.01
array/iteration/findall/bool	182572 ns	186173 ns	0.98
array/iteration/findall/int	196922 ns	195203 ns	1.01
array/iteration/findfirst/bool	117091 ns	120502 ns	0.97
array/iteration/findfirst/int	115752 ns	115651 ns	1.00
array/iteration/findmin/1d	169583 ns	169593 ns	1.00
array/iteration/findmin/2d	155532 ns	155303 ns	1.00
array/iteration/logical	353895 ns	355134 ns	1.00
array/iteration/scalar	289834 ns	289534 ns	1.00
array/permutedims/2d	74561 ns	74191 ns	1.00
array/permutedims/3d	73981 ns	73501 ns	1.01
array/permutedims/4d	77401 ns	77031 ns	1.00
array/random/rand/Float32	52890 ns	51380 ns	1.03
array/random/rand/Int64	57151 ns	57211 ns	1.00
array/random/rand!/Float32	140222 ns	90061 ns	1.56
array/random/rand!/Int64	130282 ns	93842 ns	1.39
array/random/randn/Float32	86321 ns	88552 ns	0.97
array/random/randn!/Float32	112341 ns	112572 ns	1.00
array/reductions/mapreduce/Float32/1d	132752 ns	132972 ns	1.00
array/reductions/mapreduce/Float32/dims=1	94412 ns	94771 ns	1.00
array/reductions/mapreduce/Float32/dims=1L	774191 ns	774400 ns	1.00
array/reductions/mapreduce/Float32/dims=2	96602 ns	96822 ns	1.00
array/reductions/mapreduce/Float32/dims=2L	297824 ns	298774 ns	1.00
array/reductions/mapreduce/Int64/1d	132932 ns	133662 ns	0.99
array/reductions/mapreduce/Int64/dims=1	94681 ns	95181 ns	0.99
array/reductions/mapreduce/Int64/dims=1L	784541 ns	784161 ns	1.00
array/reductions/mapreduce/Int64/dims=2	96182 ns	96381 ns	1.00
array/reductions/mapreduce/Int64/dims=2L	299884 ns	304945 ns	0.98
array/reductions/reduce/Float32/1d	132462 ns	132751 ns	1.00
array/reductions/reduce/Float32/dims=1	95112 ns	94801 ns	1.00
array/reductions/reduce/Float32/dims=1L	774601 ns	772081 ns	1.00
array/reductions/reduce/Float32/dims=2	96942 ns	94952 ns	1.02
array/reductions/reduce/Float32/dims=2L	304554 ns	305064 ns	1.00
array/reductions/reduce/Int64/1d	130412 ns	133482 ns	0.98
array/reductions/reduce/Int64/dims=1	94871 ns	94961 ns	1.00
array/reductions/reduce/Int64/dims=1L	784391 ns	782411 ns	1.00
array/reductions/reduce/Int64/dims=2	95681 ns	94081 ns	1.02
array/reductions/reduce/Int64/dims=2L	307544 ns	298415 ns	1.03
array/reverse/1d	43910 ns	44041 ns	1.00
array/reverse/1dL	75211 ns	75781 ns	0.99
array/reverse/1dL_inplace	109862 ns	169563 ns	0.65
array/reverse/1d_inplace	123012 ns	60091 ns	2.05
array/reverse/2d	51661 ns	52051 ns	0.99
array/reverse/2dL	101621 ns	101912 ns	1.00
array/reverse/2dL_inplace	126961 ns	176962 ns	0.72
array/reverse/2d_inplace	80531 ns	123882 ns	0.65
array/sorting/1d	344705 ns	339895 ns	1.01
integration/byval/reference	38710 ns	39110 ns	0.99
integration/byval/slices=1	39991 ns	39650 ns	1.01
integration/byval/slices=2	150662 ns	147182 ns	1.02
integration/byval/slices=3	247494 ns	237883 ns	1.04
integration/volumerhs	5032160 ns	5038970 ns	1.00
kernel/indexing	129262 ns	64551 ns	2.00
kernel/indexing_checked	103902 ns	59681 ns	1.74
kernel/launch	1300 ns	1290 ns	1.01
kernel/rand	85452 ns	122581 ns	0.70
latency/import	1494650233 ns	1490559052 ns	1.00
latency/precompile	12054125204 ns	11957081563 ns	1.01
latency/ttfp	10482538827 ns	10414604626 ns	1.01

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