[Do not merge] Test KernelIntrinsics

[christiangnrd]

[only tests]
[only benchmarks]

[github-actions]

Your PR requires formatting changes to meet the project's style guidelines.
Please consider running Runic (git runic master) to apply these changes.

Click here to view the suggested changes.

diff --git a/src/CUDAKernels.jl b/src/CUDAKernels.jl
index 5e39ab68e..7f8ec7ad4 100644
--- a/src/CUDAKernels.jl
+++ b/src/CUDAKernels.jl
@@ -162,29 +162,29 @@ end
 
 KI.argconvert(::CUDABackend, arg) = cudaconvert(arg)
 
-function KI.kernel_function(::CUDABackend, f::F, tt::TT=Tuple{}; name=nothing, kwargs...) where {F,TT}
+function KI.kernel_function(::CUDABackend, f::F, tt::TT = Tuple{}; name = nothing, kwargs...) where {F, TT}
     kern = cufunction(f, tt; name, kwargs...)
-    KI.Kernel{CUDABackend, typeof(kern)}(CUDABackend(), kern)
+    return KI.Kernel{CUDABackend, typeof(kern)}(CUDABackend(), kern)
 end
 
 function (obj::KI.Kernel{CUDABackend})(args...; numworkgroups = 1, workgroupsize = 1)
     KI.check_launch_args(numworkgroups, workgroupsize)
 
-    obj.kern(args...; threads=workgroupsize, blocks=numworkgroups)
+    obj.kern(args...; threads = workgroupsize, blocks = numworkgroups)
     return nothing
 end
 
 
-function KI.kernel_max_work_group_size(kernel::KI.Kernel{<:CUDABackend}; max_work_items::Int=typemax(Int))::Int
+function KI.kernel_max_work_group_size(kernel::KI.Kernel{<:CUDABackend}; max_work_items::Int = typemax(Int))::Int
     kernel_config = launch_configuration(kernel.kern.fun)
 
-    Int(min(kernel_config.threads, max_work_items))
+    return Int(min(kernel_config.threads, max_work_items))
 end
 function KI.max_work_group_size(::CUDABackend)::Int
-    Int(attribute(device(), CUDA.DEVICE_ATTRIBUTE_MAX_THREADS_PER_BLOCK))
+    return Int(attribute(device(), CUDA.DEVICE_ATTRIBUTE_MAX_THREADS_PER_BLOCK))
 end
 function KI.multiprocessor_count(::CUDABackend)::Int
-    Int(attribute(device(), CUDA.DEVICE_ATTRIBUTE_MULTIPROCESSOR_COUNT))
+    return Int(attribute(device(), CUDA.DEVICE_ATTRIBUTE_MULTIPROCESSOR_COUNT))
 end
 
 ## indexing
@@ -199,7 +199,7 @@ end
 end
 
 @device_override @inline function KI.get_global_id()
-    return (; x = Int((blockIdx().x-1)*blockDim().x + threadIdx().x), y = Int((blockIdx().y-1)*blockDim().y + threadIdx().y), z = Int((blockIdx().z-1)*blockDim().z + threadIdx().z))
+    return (; x = Int((blockIdx().x - 1) * blockDim().x + threadIdx().x), y = Int((blockIdx().y - 1) * blockDim().y + threadIdx().y), z = Int((blockIdx().z - 1) * blockDim().z + threadIdx().z))
 end
 
 @device_override @inline function KI.get_local_size()
diff --git a/src/accumulate.jl b/src/accumulate.jl
index 54fab2119..e0631d387 100644
--- a/src/accumulate.jl
+++ b/src/accumulate.jl
@@ -22,9 +22,9 @@ function partial_scan(op::Function, output::AbstractArray{T}, input::AbstractArr
     temp = CuDynamicSharedArray(T, (2*threads,))
 
     # iterate the main dimension using threads and the first block dimension
-    i = (KI.get_group_id().x-1i32) * KI.get_local_size().x + KI.get_local_id().x
+    i = (KI.get_group_id().x - 1i32) * KI.get_local_size().x + KI.get_local_id().x
     # iterate the other dimensions using the remaining block dimensions
-    j = (KI.get_group_id().z-1i32) * KI.get_num_groups().y + KI.get_group_id().y
+    j = (KI.get_group_id().z - 1i32) * KI.get_num_groups().y + KI.get_group_id().y
 
     if j > length(Rother)
         return
@@ -105,9 +105,9 @@ function aggregate_partial_scan(op::Function, output::AbstractArray,
     block = KI.get_group_id().x
 
     # iterate the main dimension using threads and the first block dimension
-    i = (KI.get_group_id().x-1i32) * KI.get_local_size().x + KI.get_local_id().x
+    i = (KI.get_group_id().x - 1i32) * KI.get_local_size().x + KI.get_local_id().x
     # iterate the other dimensions using the remaining block dimensions
-    j = (KI.get_group_id().z-1i32) * KI.get_num_groups().y + KI.get_group_id().y
+    j = (KI.get_group_id().z - 1i32) * KI.get_num_groups().y + KI.get_group_id().y
 
     @inbounds if i <= length(Rdim) && j <= length(Rother)
         I = Rother[j]
diff --git a/src/device/random.jl b/src/device/random.jl
index 7d72d90a1..063c736ed 100644
--- a/src/device/random.jl
+++ b/src/device/random.jl
@@ -73,8 +73,8 @@ end
         @inbounds global_random_counters()[warpId]
     elseif field === :ctr2
         globalId = KI.get_global_id().x +
-                   (KI.get_global_id().y - 1i32) * KI.get_global_size().x +
-                   (KI.get_global_id().z - 1i32) * KI.get_global_size().x * KI.get_global_size().y
+            (KI.get_global_id().y - 1i32) * KI.get_global_size().x +
+            (KI.get_global_id().z - 1i32) * KI.get_global_size().x * KI.get_global_size().y
         globalId%UInt32
     end::UInt32
 end
diff --git a/src/mapreduce.jl b/src/mapreduce.jl
index 97a4176b4..6fccff91e 100644
--- a/src/mapreduce.jl
+++ b/src/mapreduce.jl
@@ -294,8 +294,9 @@ function GPUArrays.mapreducedim!(f::F, op::OP, R::AnyCuArray{T},
         end
 
         partial_kernel(f, op, init, Rreduce, Rother, Val(shuffle), partial, A;
-                    threads=partial_threads, blocks=partial_blocks, shmem=partial_shmem)
-                    # workgroupsize=partial_threads, numworkgroups=partial_blocks, shmem=partial_shmem)
+            threads = partial_threads, blocks = partial_blocks, shmem = partial_shmem
+        )
+        # workgroupsize=partial_threads, numworkgroups=partial_blocks, shmem=partial_shmem)
 
         GPUArrays.mapreducedim!(identity, op, R, partial; init)
     end
diff --git a/test/base/kernelabstractions.jl b/test/base/kernelabstractions.jl
index 2f2c4300b..1e674d3be 100644
--- a/test/base/kernelabstractions.jl
+++ b/test/base/kernelabstractions.jl
@@ -4,9 +4,14 @@ using SparseArrays
 
 include(joinpath(dirname(pathof(KernelAbstractions)), "..", "test", "testsuite.jl"))
 
-Testsuite.testsuite(()->CUDABackend(false, false), "CUDA", CUDA, CuArray, CuDeviceArray; skip_tests=Set([
-    "CPU synchronization",
-    "fallback test: callable types",]))
+Testsuite.testsuite(
+    () -> CUDABackend(false, false), "CUDA", CUDA, CuArray, CuDeviceArray; skip_tests = Set(
+        [
+            "CPU synchronization",
+            "fallback test: callable types",
+        ]
+    )
+)
 for (PreferBlocks, AlwaysInline) in Iterators.product((true, false), (true, false))
     Testsuite.unittest_testsuite(()->CUDABackend(PreferBlocks, AlwaysInline), "CUDA", CUDA, CuDeviceArray)
 end
diff --git a/test/runtests.jl b/test/runtests.jl
index 802c832e5..11584ca92 100644
--- a/test/runtests.jl
+++ b/test/runtests.jl
@@ -1,6 +1,6 @@
 @static if VERSION < v"1.11" && get(ENV, "BUILDKITE_PIPELINE_NAME", "CUDA.jl") == "CUDA.jl"
     using Pkg
-    Pkg.add(url="https://github.com/JuliaGPU/KernelAbstractions.jl", rev="main")
+    Pkg.add(url = "https://github.com/JuliaGPU/KernelAbstractions.jl", rev = "main")
 end
 
 using Distributed

[github-actions]

CUDA.jl Benchmarks

Details

Benchmark suite	Current: cc9e99d	Previous: 54c7586	Ratio
array/accumulate/Float32/1d	102863 ns	100040 ns	1.03
array/accumulate/Float32/dims=1	76858 ns	75598 ns	1.02
array/accumulate/Float32/dims=1L	1612907 ns	1585923 ns	1.02
array/accumulate/Float32/dims=2	142610 ns	140396 ns	1.02
array/accumulate/Float32/dims=2L	663488 ns	652952 ns	1.02
array/accumulate/Int64/1d	120081 ns	116760 ns	1.03
array/accumulate/Int64/dims=1	81357 ns	78966 ns	1.03
array/accumulate/Int64/dims=1L	1738738 ns	1697893 ns	1.02
array/accumulate/Int64/dims=2	152875 ns	150510 ns	1.02
array/accumulate/Int64/dims=2L	961211 ns	959254 ns	1.00
array/broadcast	18714 ns	18315 ns	1.02
array/construct	1233.7 ns	1307.5 ns	0.94
array/copy	16727 ns	16659 ns	1.00
array/copyto!/cpu_to_gpu	214216 ns	211984 ns	1.01
array/copyto!/gpu_to_cpu	279232 ns	280422 ns	1.00
array/copyto!/gpu_to_gpu	10501 ns	10380 ns	1.01
array/iteration/findall/bool	135277 ns	131500 ns	1.03
array/iteration/findall/int	149390 ns	145666 ns	1.03
array/iteration/findfirst/bool	70218 ns	68920 ns	1.02
array/iteration/findfirst/int	72711 ns	71508 ns	1.02
array/iteration/findmin/1d	67591 ns	66419 ns	1.02
array/iteration/findmin/2d	101064 ns	101084 ns	1.00
array/iteration/logical	197508 ns	190475 ns	1.04
array/iteration/scalar	65391 ns	65525 ns	1.00
array/permutedims/2d	49687 ns	49765 ns	1.00
array/permutedims/3d	52205 ns	50702 ns	1.03
array/permutedims/4d	53378 ns	51020 ns	1.05
array/random/rand/Float32	11842 ns	11632 ns	1.02
array/random/rand/Int64	22838 ns	22338 ns	1.02
array/random/rand!/Float32	9865.333333333334 ns	7935.333333333333 ns	1.24
array/random/rand!/Int64	21292 ns	17826 ns	1.19
array/random/randn/Float32	36568 ns	36306 ns	1.01
array/random/randn!/Float32	24173 ns	23975 ns	1.01
array/reductions/mapreduce/Float32/1d	34927 ns	33733 ns	1.04
array/reductions/mapreduce/Float32/dims=1	41939 ns	38085 ns	1.10
array/reductions/mapreduce/Float32/dims=1L	51268 ns	50507 ns	1.02
array/reductions/mapreduce/Float32/dims=2	61266 ns	55668 ns	1.10
array/reductions/mapreduce/Float32/dims=2L	69788 ns	67497 ns	1.03
array/reductions/mapreduce/Int64/1d	43255 ns	40096 ns	1.08
array/reductions/mapreduce/Int64/dims=1	45532 ns	40849 ns	1.11
array/reductions/mapreduce/Int64/dims=1L	87025 ns	86591 ns	1.01
array/reductions/mapreduce/Int64/dims=2	65507 ns	58177 ns	1.13
array/reductions/mapreduce/Int64/dims=2L	86051 ns	83232 ns	1.03
array/reductions/reduce/Float32/1d	35080 ns	33766 ns	1.04
array/reductions/reduce/Float32/dims=1	41985 ns	38178 ns	1.10
array/reductions/reduce/Float32/dims=1L	51132 ns	50557 ns	1.01
array/reductions/reduce/Float32/dims=2	61001 ns	55534 ns	1.10
array/reductions/reduce/Float32/dims=2L	70190 ns	68199 ns	1.03
array/reductions/reduce/Int64/1d	42765 ns	39982 ns	1.07
array/reductions/reduce/Int64/dims=1	45585 ns	41049 ns	1.11
array/reductions/reduce/Int64/dims=1L	86875 ns	86662 ns	1.00
array/reductions/reduce/Int64/dims=2	65593 ns	58182 ns	1.13
array/reductions/reduce/Int64/dims=2L	85819 ns	83392 ns	1.03
array/reverse/1d	17160 ns	17096 ns	1.00
array/reverse/1dL	68165 ns	67978 ns	1.00
array/reverse/1dL_inplace	65617 ns	65357 ns	1.00
array/reverse/1d_inplace	8617.333333333334 ns	8332.666666666666 ns	1.03
array/reverse/2d	20329 ns	20040 ns	1.01
array/reverse/2dL	72257 ns	71744 ns	1.01
array/reverse/2dL_inplace	65273 ns	65037 ns	1.00
array/reverse/2d_inplace	10013 ns	9712 ns	1.03
array/sorting/1d	2715945 ns	2725346 ns	1.00
array/sorting/2d	1063388 ns	1061549 ns	1.00
array/sorting/by	3269427 ns	3268510 ns	1.00
cuda/synchronization/context/auto	1137.7 ns	1148.8 ns	0.99
cuda/synchronization/context/blocking	947.2307692307693 ns	947 ns	1.00
cuda/synchronization/context/nonblocking	6077.4 ns	6099.8 ns	1.00
cuda/synchronization/stream/auto	996.2 ns	1006 ns	0.99
cuda/synchronization/stream/blocking	802.9010989010989 ns	859.8142857142857 ns	0.93
cuda/synchronization/stream/nonblocking	5941.4 ns	5966.2 ns	1.00
integration/byval/reference	143449 ns	143152 ns	1.00
integration/byval/slices=1	145551 ns	145205 ns	1.00
integration/byval/slices=2	283846 ns	283811 ns	1.00
integration/byval/slices=3	422584 ns	422282 ns	1.00
integration/cudadevrt	101944 ns	101717 ns	1.00
integration/volumerhs	8890230 ns	8896793 ns	1.00
kernel/indexing	12860 ns	12625 ns	1.02
kernel/indexing_checked	13729 ns	13467 ns	1.02
kernel/launch	2328.222222222222 ns	2103.8888888888887 ns	1.11
kernel/occupancy	694.2013422818792 ns	692.2105263157895 ns	1.00
kernel/rand	14406 ns	14754 ns	0.98
latency/import	4017459962 ns	3848826206 ns	1.04
latency/precompile	4758697989 ns	4629101097 ns	1.03
latency/ttfp	4644072202 ns	4482824517 ns	1.04

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