Add comments

Author: jishnub

stdlib/LinearAlgebra/src/matmul.jl

@@ -935,12 +935,13 @@ Base.@constprop :aggressive generic_matmatmul!(C::AbstractVecOrMat, tA, tB, A::A
     if BxN != CxN
         throw(DimensionMismatch(lazy"matrix B has axes ($BxK,$BxN), matrix C has axes ($CxM,$CxN)"))
     end
-    _add_alpha = MulAddMul{ais1,true,typeof(_add.alpha),Bool}(_add.alpha,false)
+    _rmul_alpha = MulAddMul{ais1,true,typeof(_add.alpha),Bool}(_add.alpha,false)
     if isbitstype(R) && sizeof(R) ≤ 16 && !(A isa Adjoint || A isa Transpose)
         _rmul_or_fill!(C, _add.beta)
         (iszero(_add.alpha) || isempty(A) || isempty(B)) && return C
         @inbounds for n in BxN, k in BxK
-            Balpha = _add_alpha(B[k,n])
+            # Balpha = B[k,n] * alpha, but we skip the multiplication in case isone(alpha)
+            Balpha = _rmul_alpha(B[k,n])
             @simd for m in AxM
                 C[m,n] = muladd(A[m,k], Balpha, C[m,n])
             end