Fix 5-arg mul! for tiled generic case (#33218)

JuliaLang · Sep 11, 2019 · 09ff912 · 09ff912
1 parent 2d4f4d2
commit 09ff912
Show file tree

Hide file tree

Showing 2 changed files with 17 additions and 1 deletion.
diff --git a/stdlib/LinearAlgebra/src/matmul.jl b/stdlib/LinearAlgebra/src/matmul.jl
@@ -762,7 +762,7 @@ function _generic_matmatmul!(C::AbstractVecOrMat{R}, tA, tB, A::AbstractVecOrMat
                     if isone(_add.alpha) && iszero(_add.beta)
                         copyto!(C, ib:ilim, jb:jlim, Ctile, 1:ilen, 1:jlen)
                     else
-                        C[ib:ilim, jb:jlim] .= @views _add.(C[ib:ilim, jb:jlim], Ctile[1:ilen, 1:jlen])
+                        C[ib:ilim, jb:jlim] .= @views _add.(Ctile[1:ilen, 1:jlen], C[ib:ilim, jb:jlim])
                     end
                 end
             end

diff --git a/stdlib/LinearAlgebra/test/matmul.jl b/stdlib/LinearAlgebra/test/matmul.jl
@@ -581,4 +581,20 @@ end
     @test mul!(view(C, 1:10, 1:10), A, 0.5) == A * 0.5
 end
 
+@testset "Issue #33214: tiled generic mul!" begin
+    n = 100
+    A = rand(n, n)
+    B = rand(n, n)
+    C = zeros(n, n)
+    mul!(C, A, B, -1, 0)
+    D = -A * B
+    @test D ≈ C
+
+    # Just in case dispatching on the surface API `mul!` is changed in the future,
+    # let's test the function where the tiled multiplication is defined.
+    fill!(C, 0)
+    LinearAlgebra._generic_matmatmul!(C, 'N', 'N', A, B, LinearAlgebra.MulAddMul(-1, 0))
+    @test D ≈ C
+end
+
 end # module TestMatmul