Within the computational functionalist framework, the possibility of machine consciousness is closely tied to the manner of computational implementation. Currently, discussions on computational implementation and its constraints are increasingly becoming a prominent topic. This paper primarily examines two distinct approaches with varying levels of constraint strength: the heavy-constraint approach, represented by implementationalism, which advocates for imposing substantive constraints on computational implementation to capture the necessary conditions for consciousness; and the light-constraint approach, which proposes minimal constraints aimed to preclude computational triviality arguments, thereby committing to a mechanistic account of computational implementation. Based on a comparative analysis of these two approaches, this paper advocates for adopting the mechanistic account of computational implementation as a foundational hypothesis, upon which substantive constraints on computational implementation should be imposed.