Diffusion models have emerged as state-of-the-art generative methods, particularly excelling in conditional tasks such as prompt-driven image synthesis. While recent research emphasizes the pivotal role of noise seeds in enhancing text-image alignment and generating human-preferred outputs,these works predominantly rely on random Gaussian noise or heuristic local adjustments, , overlooking the potential of global optimization trategies to systematically improve generation quality. To bridge this gap, we propose Seed Optimization based on Evolution (SOE), a hybrid framework that integrates global evolutionary search with local semantic refinement. The global evolutionary stage conducts seed selection by jointly optimizing text-image alignment (via CLIP-Score) and human preference estimation (via ImageReward), while the local stage employs diffusion inversion to inject conditional semantics into the noise seed. Together, these components constitute a model-agnostic, training-free optimization framework for conditional diffusion models. Extensive experiments across various diffusion models demonstrate that SOE consistently improves semantic fidelity and visual quality, highlighting its generalizability and potential as a plug-and-play enhancement for generative diffusion pipelines.

In recent years, speculative execution attacks bring severe threats to hardware security. An attacker tempts the victim process to read secret data during speculative execution and recovers it via micro-architectural covert channels. Most defense strategies in the literature adopt a blocking approach, in which the execution of malicious speculative instructions is withheld until they become safe and, thus, excessive performance overhead is inevitable. Although several non-blocking approaches have also been proposed, extra hardware components are introduced to buffer data of speculative cache accesses, bringing on significant hardware costs. In this paper, we propose the Secure Access Policy (SAP) to enable the non-blocking execution of malicious speculative instructions in an isolated environment, which reuses the existing Line Fill Buffer and Translation Lookaside Buffer, without introducing any new hardware component. The isolated environment, named Safe Shelter area (SS area), maintains the data accessed by malicious speculative instructions and those data cannot be transmitted to the outside. An improved taint tracking technique is introduced to effectively reduce the number of accesses to the SS area. In addition, an extended cache coherence mechanism and a process-bound technique are also proposed to guarantee the validity and security of SS area’s data. We evaluate SAP on SPEC2006 and PARSEC3.0 workloads. The evaluation results show that SAP can effectively defend those attacks leveraging speculative executions and cache covert channels, and outperforms existing approaches in performance overhead, 2.35% and 3.21% in the Spectre and Futuristic defense models, respectively.