Current memory tiering systems mitigate asymmetric latency through page migration between tiers. While this approach effectively hides latency, it overlooks scenarios where latency could potentially be tolerated. We propose that an efficient system should integrate both latency reduction (migration) and latency tolerance strategies. Our research demonstrates the effectiveness of prefetchers in tolerating latency within such systems, highlighting their importance in the design of high-performance memory tiering solutions.

Large language models (LLMs) have demonstrated significant potential in the development of intelligent LLM-based agents. However, when users use these agent applications to perform file operations, their interaction with the file system still remains the traditional paradigm: reliant on manual navigation through precise commands. This paradigm poses a bottleneck to the usability of these systems as users are required to navigate complex folder hierarchies and remember cryptic file names. To address this limitation, we propose an LLM-based Semantic File System (LSFS) for prompt-driven file management in LLM Agent Operating System (AIOS). Unlike conventional approaches, LSFS incorporates LLMs to enable users or agents to interact with files through natural language prompts, facilitating semantic file management. At the macro-level, we develop a comprehensive API set to achieve semantic file management functionalities, such as semantic file retrieval, file update summarization, and semantic file rollback). At the micro-level, we store files by constructing semantic indexes for them, design and implement syscalls of different semantic operations, e.g., CRUD (create, read, update, delete), group by, join. Our experiments show that LSFS can achieve at least 15% retrieval accuracy improvement with 2.1× higher retrieval speed in the semantic file retrieval task compared with the traditional file system. In the traditional keyword-based file retrieval task (i.e., retrieving by string-matching), LSFS also performs stably well, i.e., over 89% F1-score with improved usability, especially when the keyword conditions become more complex. Additionally, LSFS supports more advanced file management operations, i.e., semantic file rollback and file sharing and achieves 100% success rates in these tasks, further suggesting the capability of LSFS . The code is available at https://github.com/agiresearch/AIOS-LSFS.

Page-based far memory systems transparently expand an application's memory capacity beyond a single machine without modifying application code. However, existing systems are tailored to scenarios with low application thread counts, and fail to scale on today's multi-core machines. This makes them unsuitable for data-intensive applications that both rely on far memory support and scale with increasing thread count. Our analysis reveals that this poor scalability stems from inefficient holistic coordination between page fault-in and eviction operations. As thread count increases, current systems encounter scalability bottlenecks in TLB shootdowns, page accounting, and memory allocation.

Heterogeneous defect prediction (HDP) predicts defects for the current project (target) using heterogeneous data from other projects (source), which can improve software quality effectively. The ability to reduce the distribution divergence between the source and target project is crucial to the performance of HDP. Existing HDP methods address this issue only by using feature adaptation techniques, i.e., feature matching and feature transformation. However, without considering instance adaptability differences, they cannot fully exploit the potential of instance adaptation and thus cannot make the source and target distribution completely matched. To address this problem, we propose a novel HDP approach called joint instance and feature adaptation (JIFA) combining both instance adaptation and feature adaptation to narrow the gap between projects. Making joint use of both adaptation techniques, JIFA not only further reduces the distribution discrepancy but also enhances its robustness to abnormal outliers. Besides, discriminant information is also preserved in JIFA so that a better classification boundary can be obtained. Particularly, JIFA applies to both heterogeneous and homogeneous cross-project defect prediction (CPDP) tasks. The experimental results from 22 projects indicate that JIFA outperforms a range of advanced heterogeneous and homogeneous CPDP approaches.

Heterogeneous memory systems promise better performance, energy-efficiency, and cost trade-offs in emerging systems. But delivering on this promise requires efficient OS mechanisms and policies for data tiering and migration. Unfortunately, modern OSes are lacking inefficient support for data tiering. While this problem is known for application data, the question of how best to manage kernel objects for filesystems and networking---i.e., inodes, dentry caches, journal blocks, socket buffers, etc.---has largely been ignored and presents a performance challenge for I/O-intensive workloads. We quantify the scale of this challenge and introduce a new OS abstraction, kernel-level object contexts (KLOCs), to enable efficient tiering of kernel objects. We use KLOCs to identify and group kernel objects with similar hotness, reuse, and liveness, and demonstrate their use in data placement and migration across several heterogeneous memory system configurations, including Intel’s Optane systems. Performance evaluations using RocksDB, Redis, Cassandra, and Spark show that KLOCs enable up to 2.7× higher system throughput versus prior art.