Comment by Tankenstein

Looks like a demonstration that using `long double` math requires dipping into x87 instructions, specifically the `fldt` instruction: "floating point load ten bytes".