Our mistake was following: we used wrong function of matrix multiplication. Our function had three parameters FUNC(a, b, c). We call FUNC(x, y, x) to calculate x = x * y; but there is mistake, because when we calculate resulting matrix we should use constant "x" and "y", but first source parameter "x" is the same as third resulting parameter "x". "X" is changing during calculation and so we change also source parameter.
Right decision is to use additional variable to store "x"

t = x;
FUNC(t, y, x);

That's all