Homer has, for example:
Τυδέα δ’ οὐ μέμνημαι, ἐπεί μ’ ἔτι τυτθὸν ἐόντα
κάλλιφ’, ὅτ’ ἐν Θήβῃσιν ἀπώλετο λαὸς Ἀχαιῶν. (Iliad 6.222)
But Tydeus I remember not, for he left me whilst I was yet young, when the people of the Greeks perished at Thebes. (Buckley)
I imagine that ἀπό- is an intensifier in ἀπόλλυμι, and the double omicron gets contracted to a single omicron. Wiktionary lists the spelling with omega for some tenses, not others, and lists both spellings for some tenses, e.g., for the aorist they label the omega spelling as the normal one (presumably Attic) and the omicron one as epic.
In the verb without the preposition, the principal parts listed by Pharr for the Homeric dialect are ὄλλυμι, ὀλέσω, ὤλεσα (or the same forms with -σσ-). So I would have thought that the lengthening of the omicron was actually just the normal phonetic change that you get with the augment, and that the preposition in the ἀπό- form just comes along for the ride (i.e., there is no augment happening to the alpha). But if this is the case, then why do we get both spellings in the Homeric dialect for the aorist, and also both spellings in the imperfect?
Is this a situation where many forms just aren't attested, ancient scribes were inconsistent in how they spelled things, and there were many different dialects, so the whole thing just gets fuzzy?
[EDIT] After posting this, I think I may have figured out what's going on. The augment is optional in Homer, so then we get the -ο- spelling.