Normally (in air with a non-zero viscosity), when I give a ping-pong ball top-spin, the air above the ball tends to decrease in velocity while the air on the downside tends to decrease, resulting in a downward motion, because of the Bernoulli Effect.
This is what I think:
If the viscosity (inner friction of the medium, which is air in this case) of the air is zero (hypothetical), the air molecules will affect the ball only in a thin layer that has the thickness of the maximum variation in heigth on the surface of the ball (the fact that the air has zero viscosity doesn't mean there can't be friction between the air and the ball), because of the lack of viscosity doesn't let the air outside this thin layer take part in the rotation of the air. Of course, it takes longer for the ball stop rotating, as more energy is transferred to the air with viscosity.
This results in two very thin layers on the up- and down-side of the ball where there is gonna be a tiny pressure difference (less pressure on the downside, more pressure on the upside). So the effect is much less than in air with a viscosity (unless the surface of the ball is perfectly smooth, in which case there is no difference), and the ball will deflect down a little bit.
Am I right? Or does zero-viscosity air also means that there is no friction between the air and the rough surface of the ball?