Additive manufacturing (AM) structures often suffer from significant fatigue scatter due to microstructural heterogeneity, such as porosity and crystallographic texture. To address the trade-off between computational cost and physical fidelity in high-cycle fatigue (HCF) assessment, this work presents a unified multiscale computational framework. Initially, we established a rapid fatigue limit prediction method by coupling Crystal Plasticity (CP) modeling with Direct Cyclic Analysis (DCA) and the Dang Van shakedown theory. Applied to LMD AlSi10Mg, this approach efficiently quantified the synergistic effects of defects and texture on the elastic shakedown limit [1]. Building upon this efficient steady-state solving capability, we extended the framework from the elastic shakedown domain to the alternating plasticity domain characterized by stable hysteresis loops. This extension allows for the quantification of cyclic plastic dissipation associated with stiffness degradation. Consequently, a novel Multi-Time-Scale Direct Cyclic Analysis (MTS-DCA) framework is proposed for accelerated full-life fatigue simulation. This decoupled strategy utilizes DCA as a "fast time-scale" solver to retrieve stable cyclic responses and an external numerical controller for "slow time-scale" damage integration. By incorporating non-local implicit gradient regularization, the framework effectively resolves the mesh dependency issues inherent in local damage softening. Numerical benchmarks demonstrate that the MTS-DCA method achieves a computational acceleration of nearly two orders of magnitude compared to traditional incremental analysis while maintaining high accuracy. The integrated framework not only predicts fatigue limits but also elucidates the micro-mechanisms of damage initiation and propagation, providing a robust tool for fatigue-oriented design.