I have a question regarding the performance of matrix division (/).

During some calculations, I noticed a huge impact of 3x3-Matrix divisions compared to expectations I had from a similar matlab program. Therefore, I made some test with / as well as \. I noticed significant performance improvements for A/B if I manually step through the lines of A.

Here the Code:

```
using LinearAlgebra
using BenchmarkTools
function funDiv1(A, B)
return C = A/B
end
function funDiv2(A,B)
C = similar(A)
LineNum = size(A, 1)
for iL = 1:LineNum
C[iL, :] = A[iL, :]'/B
end
return C
end
A = convert(Matrix{Float64}, [1 3 7; 17 28 33; 2 0 7])
B = convert(Matrix{Float64}, [2 5 23; 1 45 2; 4 2 12])
Bf = factorize(B)
bench1 = @benchmark funDiv1($A, $Bf)
bench2 = @benchmark funDiv2($A, $Bf)
```

On my system, the median time is about 13 micro seconds for funDiv1 and .5 for funDiv2.

Is there a good reason for the additional computation time or can I use the second function for a performance improvement?