Simulating the summation for given $A$ and $B$ to exactly hit $100$ should be straight forward, maybe except for the case that $100$ can not be reached for the given pair.
Can we say more in advance? Reaching $100$ means
$$
100 = \sum_{i=0}^n \frac{[(iA+B)\pmod {100}]}{100}
$$
This gives
$$
10000 = \sum_{i=0}^n [iA+B\pmod {100}] \quad (*)
$$
Reaching $100$:
a) To make no progress at all, we need
$$
c = i A + B \pmod {100}
$$
to vanish for all values $i$. If $c = 0$ then the linear Diophantine equation
$$
100 X - A Y = B \quad (**)
$$
has integer solutions $X$, $Y$. This requires $d=\gcd(100, A)$ to be a divisor of $B$.
We set $m = B / d$.
The equation $100 s + A t = d$ can be solved by the extended Euclidean algorithm for integers $s$ and $t$.
Then $(**)$ has the general solution $(X,Y)=(ms - (A/d) k, mt - (100/d) k)$ for $k \in \mathbb{Z}$. We need the spacing $100/d = 1$ for the $Y$ solutions, or $d = 100$. So nothing moves if $B \bmod 100 = 0$ and $\gcd(100,A) = 100$, thus $A \bmod 100 = 0$.
Estimating $n$:
Equation $(*)$ gives
\begin{align}
10000 &= \sum_{i=0}^n [iA+B\pmod {100}] \\
&= \sum_{i=0}^n iA+B - q_i 100 \\
&= \frac{n(n+1)}{2} A + (n+1) B - 100 \sum_{i=0}^n q_i \\
&= \frac{A}{2} n^2 + \left(\frac{A}{2}+B\right) n + B - 100 Q
\end{align}
for a certain $Q$ depending on $A$ and $B$ (this sum of quotients $q_i$ now contains the complication of modulo taking).
Assuming $A > 0$, solving for $n$ gives
\begin{align}
\frac{2}{A}(10000 + 100 Q - B) &=
n^2 + \left(1+\frac{2B}{A}\right) n \\
&=
\left( n + \frac{1}{2} + \frac{B}{A}\right)^2 -
\left(
\frac{1}{2} + \frac{B}{A}
\right)^2
\end{align}
and
\begin{align}
n(Q, A, B)
&=
\sqrt{\frac{2}{A}(10000 + 100 Q - B) +
\left(
\frac{1}{2} + \frac{B}{A}
\right)^2} -
\frac{1}{2} - \frac{B}{A}
\end{align}
with
$$
Q
= \sum_{i=0}^n q_i
= \sum_{i=0}^n \left\lfloor \frac{iA+B}{100} \right\rfloor
= \sum_{i=0}^n \frac{iA+B}{100} - \delta_i
= S - \sum_{i=0}^n \delta_i
$$
with $\delta_i \in [0,1)$.
Results from simulation:
a) Hitting $100$ exactly
The search space for minimal $n$ is $\{ 0, \ldots, 99 \}$ as only $A \bmod 100$ and $B \bmod 100$ matter, that cell repeats. The first smallest results $(A, B, n)$ regarding $n$ are
$$
(100, 80, 124) \\
(50, 82, 174) \\
(4, 43, 191) \\
(96, 7, 191) \\
(24, 35, 195) \\
\vdots
$$
where the first one is the same as $(0,80,124)$, but $A = 0$ was not feasible.
b) Reaching $100$ or more
In this case the fastest entries $(A,B, n)$ are
$$
(0,99,101) \\
(0,98,102) \\
(0,97,103) \\
(0,96,104) \\
(0,95,105) \\
\vdots
$$