Temporal Graph Networks (TGNs) are widely used to model evolving relationships in dynamic graphs. However, existing inference systems enforce a step-wise paradigm: processing each temporal graph sequentially with a memory update followed by aggregation. We break this dependency by decoupling memory updates from aggregation while preserving prediction accuracy, thereby enabling a global view for fine-grained parallelism control. This design unlocks new optimization opportunities but introduces three system-level challenges: managing intermediate multi-state representations, curbing memory-bound update overheads, and selecting a safe yet efficient aggregation granularity. We present \textit{APERTURE}, a TGN inference framework that bridges algorithmic semantics and system design. To address the above challenges, \textit{APERTURE} (1) jointly aggregates temporal states via computation graph transformation, (2) minimizes redundant memory traffic through dependency-aware update reconstruction; (3) selects the optimal granularity by analytically modeling. The experimental results show that \textit{APERTURE} achieves up to 59.3$\times$ speedup over state-of-the-art baselines without compromising accuracy.