Allow for ndims query

[vchuravy]

No description provided.

[github-actions]

Benchmark Results

	main	8671ed9...	main/8671ed9d6b1358...
saxpy/default/Float16/1024	0.734 ± 0.0075 μs	0.723 ± 0.01 μs	1.01
saxpy/default/Float16/1048576	0.173 ± 0.0094 ms	0.173 ± 0.0095 ms	1
saxpy/default/Float16/16384	3.36 ± 0.052 μs	3.32 ± 0.023 μs	1.01
saxpy/default/Float16/2048	0.912 ± 0.012 μs	0.906 ± 0.012 μs	1.01
saxpy/default/Float16/256	0.59 ± 0.0058 μs	0.573 ± 0.0076 μs	1.03
saxpy/default/Float16/262144	0.044 ± 0.00053 ms	0.0439 ± 0.00081 ms	1
saxpy/default/Float16/32768	6.05 ± 0.1 μs	5.99 ± 0.05 μs	1.01
saxpy/default/Float16/4096	1.31 ± 0.023 μs	1.3 ± 0.023 μs	1.01
saxpy/default/Float16/512	0.648 ± 0.007 μs	0.632 ± 0.0076 μs	1.02
saxpy/default/Float16/64	0.556 ± 0.0062 μs	0.539 ± 0.0058 μs	1.03
saxpy/default/Float16/65536	11.7 ± 0.2 μs	11.7 ± 0.13 μs	1
saxpy/default/Float32/1024	0.632 ± 0.011 μs	0.63 ± 0.01 μs	1
saxpy/default/Float32/1048576	0.231 ± 0.012 ms	0.238 ± 0.018 ms	0.969
saxpy/default/Float32/16384	3.49 ± 1 μs	2.74 ± 0.15 μs	1.28
saxpy/default/Float32/2048	0.745 ± 0.015 μs	0.74 ± 0.053 μs	1.01
saxpy/default/Float32/256	0.559 ± 0.0066 μs	0.556 ± 0.0077 μs	1.01
saxpy/default/Float32/262144	0.0451 ± 0.0048 ms	0.0533 ± 0.013 ms	0.846
saxpy/default/Float32/32768	6.65 ± 1.5 μs	5.24 ± 0.32 μs	1.27
saxpy/default/Float32/4096	1.12 ± 0.068 μs	1.12 ± 0.083 μs	1
saxpy/default/Float32/512	0.597 ± 0.0081 μs	0.595 ± 0.0098 μs	1
saxpy/default/Float32/64	0.55 ± 0.0064 μs	0.548 ± 0.0066 μs	1
saxpy/default/Float32/65536	13 ± 1.4 μs	12.3 ± 0.69 μs	1.06
saxpy/default/Float64/1024	0.755 ± 0.017 μs	0.749 ± 0.063 μs	1.01
saxpy/default/Float64/1048576	0.478 ± 0.051 ms	0.48 ± 0.048 ms	0.996
saxpy/default/Float64/16384	6.25 ± 1.5 μs	5.24 ± 0.28 μs	1.19
saxpy/default/Float64/2048	1.13 ± 0.067 μs	1.13 ± 0.085 μs	1
saxpy/default/Float64/256	0.589 ± 0.0082 μs	0.574 ± 0.0081 μs	1.03
saxpy/default/Float64/262144	0.0902 ± 0.0093 ms	0.0891 ± 0.0099 ms	1.01
saxpy/default/Float64/32768	12.8 ± 1.3 μs	11.9 ± 0.94 μs	1.07
saxpy/default/Float64/4096	1.88 ± 0.25 μs	1.67 ± 0.11 μs	1.12
saxpy/default/Float64/512	0.642 ± 0.01 μs	0.624 ± 0.01 μs	1.03
saxpy/default/Float64/64	0.564 ± 0.0069 μs	0.549 ± 0.0072 μs	1.03
saxpy/default/Float64/65536	24.1 ± 2.6 μs	22.8 ± 1.9 μs	1.06
saxpy/static workgroup=(1024,)/Float16/1024	2.18 ± 0.03 μs	2.16 ± 0.029 μs	1.01
saxpy/static workgroup=(1024,)/Float16/1048576	0.157 ± 0.0084 ms	0.159 ± 0.0093 ms	0.99
saxpy/static workgroup=(1024,)/Float16/16384	4.41 ± 0.1 μs	4.4 ± 0.11 μs	1
saxpy/static workgroup=(1024,)/Float16/2048	2.34 ± 0.029 μs	2.32 ± 0.029 μs	1.01
saxpy/static workgroup=(1024,)/Float16/256	2.81 ± 0.04 μs	2.8 ± 0.039 μs	1
saxpy/static workgroup=(1024,)/Float16/262144	0.0418 ± 0.0016 ms	0.0419 ± 0.0013 ms	0.999
saxpy/static workgroup=(1024,)/Float16/32768	6.8 ± 0.16 μs	6.84 ± 0.16 μs	0.994
saxpy/static workgroup=(1024,)/Float16/4096	2.66 ± 0.037 μs	2.66 ± 0.046 μs	1
saxpy/static workgroup=(1024,)/Float16/512	3.25 ± 0.048 μs	3.24 ± 0.039 μs	1
saxpy/static workgroup=(1024,)/Float16/64	2.51 ± 0.21 μs	2.49 ± 0.21 μs	1.01
saxpy/static workgroup=(1024,)/Float16/65536	12.4 ± 0.28 μs	12.3 ± 0.2 μs	1
saxpy/static workgroup=(1024,)/Float32/1024	2.22 ± 0.032 μs	2.23 ± 0.037 μs	0.995
saxpy/static workgroup=(1024,)/Float32/1048576	0.207 ± 0.027 ms	0.206 ± 0.038 ms	1.01
saxpy/static workgroup=(1024,)/Float32/16384	4.37 ± 0.26 μs	4.4 ± 0.22 μs	0.995
saxpy/static workgroup=(1024,)/Float32/2048	2.37 ± 0.055 μs	2.39 ± 0.049 μs	0.994
saxpy/static workgroup=(1024,)/Float32/256	2.66 ± 0.043 μs	2.67 ± 0.078 μs	0.998
saxpy/static workgroup=(1024,)/Float32/262144	0.048 ± 0.0039 ms	0.0478 ± 0.0039 ms	1
saxpy/static workgroup=(1024,)/Float32/32768	7.49 ± 0.39 μs	7.59 ± 0.32 μs	0.987
saxpy/static workgroup=(1024,)/Float32/4096	2.64 ± 0.066 μs	2.67 ± 0.086 μs	0.986
saxpy/static workgroup=(1024,)/Float32/512	2.7 ± 0.073 μs	2.71 ± 0.095 μs	0.999
saxpy/static workgroup=(1024,)/Float32/64	2.7 ± 5 μs	2.71 ± 5.1 μs	0.998
saxpy/static workgroup=(1024,)/Float32/65536	14.4 ± 1.4 μs	14.6 ± 1.4 μs	0.986
saxpy/static workgroup=(1024,)/Float64/1024	2.34 ± 0.065 μs	2.31 ± 0.055 μs	1.01
saxpy/static workgroup=(1024,)/Float64/1048576	0.499 ± 0.052 ms	0.538 ± 0.065 ms	0.928
saxpy/static workgroup=(1024,)/Float64/16384	7.47 ± 0.42 μs	7.5 ± 0.38 μs	0.996
saxpy/static workgroup=(1024,)/Float64/2048	2.62 ± 0.079 μs	2.6 ± 0.068 μs	1.01
saxpy/static workgroup=(1024,)/Float64/256	2.68 ± 0.073 μs	2.66 ± 0.076 μs	1.01
saxpy/static workgroup=(1024,)/Float64/262144	0.0988 ± 0.011 ms	0.119 ± 0.0079 ms	0.83
saxpy/static workgroup=(1024,)/Float64/32768	14.5 ± 1.5 μs	15.4 ± 1.1 μs	0.937
saxpy/static workgroup=(1024,)/Float64/4096	3.18 ± 0.17 μs	3.14 ± 0.14 μs	1.01
saxpy/static workgroup=(1024,)/Float64/512	2.67 ± 0.065 μs	2.65 ± 0.068 μs	1.01
saxpy/static workgroup=(1024,)/Float64/64	2.64 ± 4.4 μs	2.61 ± 0.064 μs	1.01
saxpy/static workgroup=(1024,)/Float64/65536	26.5 ± 2.3 μs	31.5 ± 1.6 μs	0.84
time_to_load	0.322 ± 0.0018 s	0.322 ± 0.004 s	1

Benchmark Plots

A plot of the benchmark results have been uploaded as an artifact to the workflow run for this PR.
Go to "Actions"->"Benchmark a pull request"->[the most recent run]->"Artifacts" (at the bottom).

[vchuravy]

@maleadt the idea would be that backends could write in their index functions:

if ndims(ctx) <= 3
    # Do custom indexing using hardware indexes
else
   # Cartesian mess
end

But this fails if workgroups are bigger than what the launch configuration supports.
I think I will merge this nonetheless since it is just an exposing of the information.

[vchuravy]

@fredrikekre :

[1920] signal 11 (1): Segmentation fault
in expression starting at none:0
verified_kids at /home/runner/.julia/packages/Runic/pKiKZ/src/Runic.jl:95 [inlined]
second_last_leaf at /home/runner/.julia/packages/Runic/pKiKZ/src/chisels.jl:428
second_last_leaf at /home/runner/.julia/packages/Runic/pKiKZ/src/chisels.jl:430 [inlined]
second_last_leaf at /home/runner/.julia/packages/Runic/pKiKZ/src/chisels.jl:430 [inlined]
second_last_leaf at /home/runner/.julia/packages/Runic/pKiKZ/src/chisels.jl:420
no_leading_and_single_trailing_newline at /home/runner/.julia/packages/Runic/pKiKZ/src/runestone.jl:1626
format_node! at /home/runner/.julia/packages/Runic/pKiKZ/src/Runic.jl:421
format_tree! at /home/runner/.julia/packages/Runic/pKiKZ/src/Runic.jl:466
main at /home/runner/.julia/packages/Runic/pKiKZ/src/main.jl:248
jfptr_main_693 at /home/runner/.julia/compiled/v1.12/Runic/xUEjj_6UK4J.so (unknown line)
jl_apply at /cache/build/tester-demeter6-10/julialang/julia-master/src/julia.h:2250 [inlined]
jl_f__call_latest at /cache/build/tester-demeter6-10/julialang/julia-master/src/builtins.c:883
#invokelatest#1 at ./essentials.jl:1056 [inlined]
invokelatest at ./essentials.jl:1052 [inlined]
_start at ./client.jl:563
jfptr__start_57101.1 at /opt/hostedtoolcache/julia/nightly/x64/lib/julia/sys.so (unknown line)
jl_apply at /cache/build/tester-demeter6-10/julialang/julia-master/src/julia.h:2250 [inlined]
true_main at /cache/build/tester-demeter6-10/julialang/julia-master/src/jlapi.c:922
jl_repl_entrypoint at /cache/build/tester-demeter6-10/julialang/julia-master/src/jlapi.c:1081
main at julia (unknown line)
unknown function (ip: 0x7f442762a1c9) at /lib/x86_64-linux-gnu/libc.so.6
__libc_start_main at /lib/x86_64-linux-gnu/libc.so.6 (unknown line)
unknown function (ip: 0x4010b8) at julia
Allocations: 1 (Pool: 1; Big: 0); GC: 0
xargs: julia: terminated by signal 11

[github-actions]

Runic suggested the following formatting changes.

[ChrisRackauckas]

arrrrrrrr! Attack of the pirates!!! This is causing a lot of downstream breakage: can it not dispatch on ctx being a context object?