What's the point of using rcp() in shaders?

Unity Engine
, , , , ,
1

Reading through shader code in URP, I noticed that in many places divisions are written as

float result = dividend * rcp(divisor);

I understand that this is supposedly faster:

rcp

Calculates a fast, approximate, per-component reciprocal.

However, in some quick tests, I cannot see any difference in the compiled code.

fixed4 frag(v2f i) : SV_Target
{
	float a = rcp(i.a.x);
	return fixed4(a, 1, 2, 3);
}

fixed4 frag(v2f i) : SV_Target
{
	float a = 1.0 / i.a.x;
	return fixed4(a, 1, 2, 3);
}

Both of the above compiles to the following:

      ps_4_0
      dcl_input_ps linear v1.x
      dcl_output o0.xyzw
   0: div o0.x, l(1.000000, 1.000000, 1.000000, 1.000000), v1.x
   1: mov o0.yzw, l(0,1.000000,2.000000,3.000000)
   2: ret

Under what circumstances does rcp make a difference?