Merge pull request #27 from YingboMa/myb/1.7

1.7 compat

Author: YingboMa

.github/workflows/ci.yml

@@ -8,12 +8,16 @@ on:
       - master
 jobs:
   test:
-    runs-on: ubuntu-latest
+    runs-on: ${{ matrix.os }}
+    strategy:
+      matrix:
+        julia-version: ['1', '^1.7.0-0']
+        os: [ubuntu-latest, windows-latest, macOS-latest]
     steps:
       - uses: actions/checkout@v2
       - uses: julia-actions/setup-julia@v1
         with:
-          version: 1
+          version: ${{ matrix.julia-version }}
       - uses: actions/cache@v1
         env:
           cache-name: cache-artifacts

Project.toml

@@ -1,7 +1,7 @@
 name = "RecursiveFactorization"
 uuid = "f2c3362d-daeb-58d1-803e-2bc74f2840b4"
 authors = ["Yingbo Ma <[email protected]>"]
-version = "0.1.12"
+version = "0.1.13"
 
 [deps]
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"

src/lu.jl

@@ -1,17 +1,27 @@
 using LoopVectorization
 using LinearAlgebra: BlasInt, BlasFloat, LU, UnitLowerTriangular, ldiv!, checknonsingular, BLAS, LinearAlgebra
 
-function lu(A::AbstractMatrix, pivot::Union{Val{false}, Val{true}} = Val(true); kwargs...)
-    return lu!(copy(A), pivot; kwargs...)
+# 1.7 compat
+normalize_pivot(t::Val{T}) where T = t
+to_stdlib_pivot(t) = t
+if VERSION >= v"1.7.0-DEV.1188"
+    normalize_pivot(::LinearAlgebra.RowMaximum) = Val(true)
+    normalize_pivot(::LinearAlgebra.NoPivot) = Val(false)
+    to_stdlib_pivot(::Val{true}) = LinearAlgebra.RowMaximum()
+    to_stdlib_pivot(::Val{false}) = LinearAlgebra.NoPivot()
 end
 
-function lu!(A, pivot::Union{Val{false}, Val{true}} = Val(true); check=true, kwargs...)
+function lu(A::AbstractMatrix, pivot = Val(true); kwargs...)
+    return lu!(copy(A), normalize_pivot(pivot); kwargs...)
+end
+
+function lu!(A, pivot = Val(true); check=true, kwargs...)
     m, n  = size(A)
     minmn = min(m, n)
     F = if minmn < 10 # avx introduces small performance degradation
-        LinearAlgebra.generic_lufact!(A, pivot; check=check)
+        LinearAlgebra.generic_lufact!(A, to_stdlib_pivot(pivot); check=check)
     else
-        lu!(A, Vector{BlasInt}(undef, minmn), pivot; check=check, kwargs...)
+        lu!(A, Vector{BlasInt}(undef, minmn), normalize_pivot(pivot); check=check, kwargs...)
     end
     return F
 end
@@ -20,21 +30,28 @@ const RECURSION_THRESHOLD = Ref(-1)
 
 # AVX512 needs a smaller recursion limit
 function pick_threshold()
-    blasvendor = BLAS.vendor()
     RECURSION_THRESHOLD[] >= 0 && return RECURSION_THRESHOLD[]
+    blasvendor = @static if VERSION >= v"1.7.0-DEV.610"
+        :openblas64
+    else
+        BLAS.vendor()
+    end
     if blasvendor === :openblas || blasvendor === :openblas64
         LoopVectorization.register_size() == 64 ? 110 : 72
     else
         LoopVectorization.register_size() == 64 ? 48 : 72
     end
 end
 
-function lu!(A::AbstractMatrix{T}, ipiv::AbstractVector{<:Integer},
-             pivot::Union{Val{false}, Val{true}} = Val(true);
-             check::Bool=true,
-             # the performance is not sensitive wrt blocksize, and 16 is a good default
-             blocksize::Integer=16,
-             threshold::Integer=pick_threshold()) where T
+function lu!(
+    A::AbstractMatrix{T}, ipiv::AbstractVector{<:Integer},
+    pivot = Val(true);
+    check::Bool=true,
+    # the performance is not sensitive wrt blocksize, and 16 is a good default
+    blocksize::Integer=16,
+    threshold::Integer=pick_threshold()
+) where T
+    pivot = normalize_pivot(pivot)
     info = zero(BlasInt)
     m, n = size(A)
     mnmin = min(m, n)
@@ -187,7 +204,7 @@ function _generic_lufact!(A, ::Val{Pivot}, ipiv, info) where Pivot
             elseif info == 0
                 info = k
             end
-            k == minmn && break 
+            k == minmn && break
             # Update the rest
             @avx for j = k+1:n
                 for i = k+1:m

test/runtests.jl

@@ -18,7 +18,8 @@ end
 @testset "Test LU factorization" begin
     for _p in (true, false), T in (Float64, Float32, ComplexF64, ComplexF32, Real)
         p = Val(_p)
-        for s in [1:10; 50:80:200; 300]
+        for (i, s) in enumerate([1:10; 50:80:200; 300])
+            iseven(i) && (p = RecursiveFactorization.to_stdlib_pivot(p))
             siz = (s, s+2)
             @info("size: $(siz[1]) × $(siz[2]), T = $T, p = $_p")
             if isconcretetype(T)