Is it possible that an incremental work may get selected by a top conference?

It's possible, in the sense that there's a non-zero chance of acceptance. A couple of factors strongly contribute to this chance: How well can you explain the significance of the problem? How big is the novelty delta, for example, are there some entirely new concepts? How big is the runtime improvement? How rigorous is the evaluation?

In essence, you need to write a very strong paper, which is naturally easier for some contributions than for others.

If the work is significant and a good contribution, then it should get selected to a related good conference. The selection also matters if the conference you are submitting is related to the area you have contributed.

I ask my supervisor he said it seems ok to him

What do you mean by ok? If he agreed to submit to the good conference then go ahead and give a try.

Nothing wrong in giving a try to submit it to a top conference and if you are unlucky then submit it to a lower grade conference.