diff --git a/src/matmul.jl b/src/matmul.jl
index 3ad84cc1..21846b96 100644
--- a/src/matmul.jl
+++ b/src/matmul.jl
@@ -130,7 +130,19 @@ LinearAlgebra.mul!(C::AbstractVecOrMat{<:RealOrComplexI}, A::AbstractMatrix, B::
 LinearAlgebra.mul!(C::AbstractVecOrMat{<:RealOrComplexI}, A::AbstractMatrix{<:RealOrComplexI}, B::AbstractVecOrMat, α::Number, β::Number) =
     _mul!(matmul_mode(), C, A, B, α, β)
 
-_mul!(::MatMulMode{:slow}, C, A, B, α, β) = LinearAlgebra._mul!(C, A, B, α, β)
+function _mul!(::MatMulMode{:slow}, C, A::AbstractMatrix, B::AbstractVector, α, β)
+    @inbounds for l ∈ axes(A, 2), i ∈ axes(A, 1)
+        C[i] = A[i,l] * B[l] * α + C[i] * β
+    end
+    return C
+end
+
+function _mul!(::MatMulMode{:slow}, C, A::AbstractMatrix, B::AbstractMatrix, α, β)
+    @inbounds for j ∈ axes(B, 2), l ∈ axes(A, 2), i ∈ axes(A, 1)
+        C[i,j] = A[i,l] * B[l,j] * α + C[i,j] * β
+    end
+    return C
+end
 
 # fast matrix multiplication
 # Note: Rump's algorithm