Test ZeroInitializedStridedPointer.

Project.toml

@@ -18,7 +18,7 @@ OffsetArrays = "1"
 SIMDPirates = "0.8.20"
 SLEEFPirates = "0.5.4"
 UnPack = "0,1"
-VectorizationBase = "0.12.29"
+VectorizationBase = "0.12.30"
 julia = "1.1"
 
 [extras]

benchmark/Manifest.toml

@@ -154,9 +154,9 @@ version = "1.0.1"
 
 [[Distances]]
 deps = ["LinearAlgebra", "Statistics"]
-git-tree-sha1 = "bed62cc5afcff16de797a9f38fb358b74071f785"
+git-tree-sha1 = "23717536c81b63e250f682b0e0933769eecd1411"
 uuid = "b4f34e82-e78d-54a5-968a-f98e89d6e8f7"
-version = "0.9.0"
+version = "0.8.2"
 
 [[Distributed]]
 deps = ["Random", "Serialization", "Sockets"]
@@ -343,9 +343,9 @@ version = "0.21.0"
 
 [[Juno]]
 deps = ["Base64", "Logging", "Media", "Profile"]
-git-tree-sha1 = "a686b0cf235fa3e491b79b4783c2d2382292b436"
+git-tree-sha1 = "90976c3ab792a98d240d42f9df07420ccfc60668"
 uuid = "e5e0dc1b-0480-54bc-9374-aad01c23163d"
-version = "0.8.2"
+version = "0.8.3"
 
 [[KernelDensity]]
 deps = ["Distributions", "FFTW", "Interpolations", "Optim", "StatsBase", "Test"]
@@ -413,17 +413,17 @@ deps = ["Libdl"]
 uuid = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
 
 [[Loess]]
-deps = ["Distances", "Random", "Statistics", "Test"]
-git-tree-sha1 = "0ee46caf683a422b595be4dfaed6cda28f541e25"
+deps = ["Distances", "Statistics"]
+git-tree-sha1 = "67b1bc5f42144fb83b065b8a8e117e56ec10e601"
 uuid = "4345ca2d-374a-55d4-8d30-97f9976e7612"
-version = "0.5.0"
+version = "0.5.1"
 
 [[Logging]]
 uuid = "56ddb016-857b-54e1-b83d-db4d58db5568"
 
 [[LoopVectorization]]
 deps = ["DocStringExtensions", "LinearAlgebra", "OffsetArrays", "SIMDPirates", "SLEEFPirates", "UnPack", "VectorizationBase"]
-path = "/home/chriselrod/.julia/dev/LoopVectorization"
+git-tree-sha1 = "695da000214847a1489c8d78ef781e0abd5b944a"
 uuid = "bdcacae8-1622-11e9-2a5c-532679323890"
 version = "0.8.21"
 

test/gemm.jl

@@ -631,7 +631,19 @@
     LoopVectorization.maybestaticsize(::LinearAlgebra.Adjoint{T,SizedMatrix{M,N,T}}, ::Val{2}) where {M,N,T} = LoopVectorization.Static{M}()
     LoopVectorization.maybestaticsize(::LinearAlgebra.Transpose{T,SizedMatrix{M,N,T}}, ::Val{1}) where {M,N,T} = LoopVectorization.Static{N}()
     LoopVectorization.maybestaticsize(::LinearAlgebra.Transpose{T,SizedMatrix{M,N,T}}, ::Val{2}) where {M,N,T} = LoopVectorization.Static{M}()
-
+
+struct ZeroInitializedArray{T,N,A<:DenseArray{T,N}} <: DenseArray{T,N}
+    data::A
+end
+Base.size(A::ZeroInitializedArray) = size(A.data)
+Base.length(A::ZeroInitializedArray) = length(A.data)
+Base.axes(A::ZeroInitializedArray, i) = axes(A.data, i)
+@inline Base.getindex(A::ZeroInitializedArray{T}) where {T} = zero(T)
+Base.@propagate_inbounds Base.setindex!(A::ZeroInitializedArray, v, i...) = setindex!(A.data, v, i...)
+function LoopVectorization.VectorizationBase.stridedpointer(A::ZeroInitializedArray)
+    LoopVectorization.VectorizationBase.ZeroInitializedStridedPointer(LoopVectorization.VectorizationBase.stridedpointer(A.data))
+end
+
     for T ∈ (Float32, Float64, Int32, Int64)
     # let T = Int32
         # exceeds_time_limit() && break
@@ -684,6 +696,8 @@
             @test C ≈ C2
             AmuladdBavx!(C, At', Bt', 2, -1)
             @test C ≈ C2
+            fill!(C, 9999.999); AmuladdBavx!(ZeroInitializedArray(C), At', Bt')
+            @test C ≈ C2
             fill!(C, 9999.999); AmulB2x2avx!(C, A, B);
             @test C ≈ C2
             fill!(C, 9999.999); AmulB2x2avx!(C, At', B);