This is impossible for me.

dx = 192 - 147 = 45 | t = 20 | 36 * dx / t = 81 <--
dx = 1.920000 - 1.470000 = 0.450000 | dx * (3600.0/t) = 81.000000 <--
Expected: equal to 80 // WTF?
Actual: 81

I tried using floats with different accuracy, even tried adding constants to make it worse and it always somewhere fails.
IHMO the Test Cases are really bad. Someone should check this.