Momentum Overtakes the Minimum

The loss $L(w) = 2w^2 - 8w$ is minimised twice, both runs starting from $w_0 = 5$ with $\alpha = 0.1$ .

Run A — plain gradient descent
$w_{t+1} = w_t - \alpha\,L'(w_t)$

Run B — momentum (velocity starts at zero, $\beta = 0.9$ )
$v_t = \beta\,v_{t-1} + L'(w_t), \qquad w_{t+1} = w_t - \alpha\,v_t$

Run each for three updates. What is $w_3$ in each run, and what has momentum actually done?