Adapt to KernelIntrinsics

[christiangnrd]

No description provided.

[christiangnrd]

id for AMDGPU is used to identify whether we need to launch hostcalls, based on the specific id of the global variable. It is not needed for KA, so can be dropped as in https://github.com/JuliaGPU/AMDGPU.jl/pull/868/files#diff-082b94339c8f038178ee472ca9b6feec6f27f434c469138f168031f248f223f9R197.

(From JuliaGPU/KernelAbstractions.jl#666)

Is my changing static local code generation from LLVMExternalLinkage to LLVMInternalLinkage going to cause any unintended effects? I had to do so to prevent local memory allocations from being treated as the same one if there happen to be more than one in a kernel. Otherwise I think we’ll have to readd the id parameter.

[pxl-th]

Is my changing static local code generation from LLVMExternalLinkage to LLVMInternalLinkage going to cause any unintended effects?

I don't think. I think it is a remnant from pre-HIP times.

[github-actions]

AMDGPU.jl Benchmarks

Details

Benchmark suite	Current: ac192c6	Previous: 756602c	Ratio
amdgpu/synchronization/context/device	590 ns	600 ns	0.98
amdgpu/synchronization/stream/blocking	250 ns	240 ns	1.04
amdgpu/synchronization/stream/nonblocking	330 ns	340 ns	0.97
array/accumulate/Float32/1d	86881 ns	86251 ns	1.01
array/accumulate/Float32/dims=1	391965 ns	393845 ns	1.00
array/accumulate/Float32/dims=1L	134921 ns	131681 ns	1.02
array/accumulate/Float32/dims=2	131872 ns	103022 ns	1.28
array/accumulate/Float32/dims=2L	2698909 ns	2827930 ns	0.95
array/accumulate/Int64/1d	94831 ns	96412 ns	0.98
array/accumulate/Int64/dims=1	284074 ns	285244 ns	1.00
array/accumulate/Int64/dims=1L	170523 ns	160812 ns	1.06
array/accumulate/Int64/dims=2	125882 ns	120772 ns	1.04
array/accumulate/Int64/dims=2L	2927692 ns	3014433 ns	0.97
array/broadcast	117332 ns	128932 ns	0.91
array/construct	1700 ns	1680 ns	1.01
array/copy	37031 ns	39371 ns	0.94
array/copyto!/cpu_to_gpu	120822 ns	114832 ns	1.05
array/copyto!/gpu_to_cpu	124632 ns	152432 ns	0.82
array/copyto!/gpu_to_gpu	98461 ns	88321 ns	1.11
array/iteration/findall/bool	180012 ns	181912 ns	0.99
array/iteration/findall/int	192042 ns	190933 ns	1.01
array/iteration/findfirst/bool	119882 ns	114451 ns	1.05
array/iteration/findfirst/int	115222 ns	116331 ns	0.99
array/iteration/findmin/1d	168752 ns	166203 ns	1.02
array/iteration/findmin/2d	157093 ns	156173 ns	1.01
array/iteration/logical	355825 ns	346025 ns	1.03
array/iteration/scalar	289824 ns	289864 ns	1.00
array/permutedims/2d	75511 ns	64761 ns	1.17
array/permutedims/3d	76151 ns	73791 ns	1.03
array/permutedims/4d	86581 ns	76481 ns	1.13
array/random/rand/Float32	52520 ns	51540 ns	1.02
array/random/rand/Int64	60340 ns	56210 ns	1.07
array/random/rand!/Float32	93992 ns	142162 ns	0.66
array/random/rand!/Int64	124721 ns	141832 ns	0.88
array/random/randn/Float32	102982 ns	86921 ns	1.18
array/random/randn!/Float32	109462 ns	152202 ns	0.72
array/reductions/mapreduce/Float32/1d	131772 ns	132902 ns	0.99
array/reductions/mapreduce/Float32/dims=1	94731 ns	95052 ns	1.00
array/reductions/mapreduce/Float32/dims=1L	774772 ns	777081 ns	1.00
array/reductions/mapreduce/Float32/dims=2	96521 ns	96731 ns	1.00
array/reductions/mapreduce/Float32/dims=2L	302164 ns	299584 ns	1.01
array/reductions/mapreduce/Int64/1d	132812 ns	133322 ns	1.00
array/reductions/mapreduce/Int64/dims=1	94811 ns	78081 ns	1.21
array/reductions/mapreduce/Int64/dims=1L	780651 ns	783471 ns	1.00
array/reductions/mapreduce/Int64/dims=2	95942 ns	96252 ns	1.00
array/reductions/mapreduce/Int64/dims=2L	300344 ns	308254 ns	0.97
array/reductions/reduce/Float32/1d	132252 ns	132802 ns	1.00
array/reductions/reduce/Float32/dims=1	94301 ns	94832 ns	0.99
array/reductions/reduce/Float32/dims=1L	774352 ns	774621 ns	1.00
array/reductions/reduce/Float32/dims=2	96151 ns	96802 ns	0.99
array/reductions/reduce/Float32/dims=2L	295874 ns	307245 ns	0.96
array/reductions/reduce/Int64/1d	132822 ns	129672 ns	1.02
array/reductions/reduce/Int64/dims=1	94751 ns	78151 ns	1.21
array/reductions/reduce/Int64/dims=1L	781571 ns	781931 ns	1.00
array/reductions/reduce/Int64/dims=2	95961 ns	96192 ns	1.00
array/reductions/reduce/Int64/dims=2L	306864 ns	298414 ns	1.03
array/reverse/1d	43951 ns	44380 ns	0.99
array/reverse/1dL	75331 ns	74131 ns	1.02
array/reverse/1dL_inplace	114591 ns	108282 ns	1.06
array/reverse/1d_inplace	78351 ns	86471 ns	0.91
array/reverse/2d	51401 ns	50661 ns	1.01
array/reverse/2dL	100962 ns	100341 ns	1.01
array/reverse/2dL_inplace	126372 ns	117622 ns	1.07
array/reverse/2d_inplace	72271 ns	95391 ns	0.76
array/sorting/1d	342675 ns	341945 ns	1.00
integration/byval/reference	38860 ns	38830 ns	1.00
integration/byval/slices=1	39931 ns	40880 ns	0.98
integration/byval/slices=2	135882 ns	158462 ns	0.86
integration/byval/slices=3	239904 ns	238013 ns	1.01
integration/volumerhs	5044422 ns	4942659 ns	1.02
kernel/indexing	54831 ns	43630 ns	1.26
kernel/indexing_checked	63931 ns	128022 ns	0.50
kernel/launch	1310 ns	1290 ns	1.02
kernel/rand	197883 ns	106671 ns	1.86
latency/import	1570558951 ns	1501349912 ns	1.05
latency/precompile	11972948400 ns	12041117438 ns	0.99
latency/ttfp	13963592706 ns	10491950084 ns	1.33

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