eZNS - An Elastic Zoned Namespace for Commodity ZNS SSDs
Innovative storage solution enhancing cost-efficiency and performance of SSDs through adaptive zoning, offering improved throughput and reduced latency for applications like RocksDB.
Karma - Resource Allocation for Dynamic Demands
Explores Karma, a novel resource allocation algorithm extending max-min fairness for dynamic user demands, ensuring efficiency, strategy-proofness, and fairness in systems with fluctuating resource needs.
TAILCHECK - A Lightweight Heap Overflow Detection Mechanism with Page Protection and Tagged Pointers
Innovative heap overflow detection technique using page protection and pointer tagging, offering low-cost security for C/C++ programs with minimal performance impact.
Conveyor - One-Tool-Fits-All Continuous Software Deployment at Meta
Explore Meta's Conveyor, a universal software deployment tool managing 30,000+ pipelines across millions of machines, featuring automated deployments, safe updates, and ML model deployment capabilities.
LVMT - An Efficient Authenticated Storage for Blockchain
Explore LVMT, a novel blockchain storage solution that significantly improves performance by reducing I/O amplifications and enhancing authenticated storage access efficiency.
BWoS - Formally Verified Block-based Work Stealing for Parallel Processing
Innovative block-based work stealing technique for parallel processing, improving performance and reducing interference in multicore systems. Formally verified and optimized for weak memory models.
Accelerating Graph Neural Networks with Fine-Grained Intra-Kernel Communication-Computation Pipelining on Multi-GPU Platforms
Innovative system design accelerating graph neural networks on multi-GPU platforms through fine-grained computation-communication pipelining, outperforming existing frameworks.
Johnny Cache - The End of DRAM Cache Conflicts in Tiered Main Memory Systems
Innovative hardware management techniques for tiered memory systems, demonstrating superior performance by eliminating cache conflicts through low-overhead methods at page allocation time.
Encrypted Databases Made Secure Yet Maintainable
Introduces HEDB, a novel encrypted database system that prevents smuggle attacks while maintaining database functionality. Balances security and maintainability through dual-mode design.
ScaleDB - A Scalable, Asynchronous In-Memory Database
Explore ScaleDB, a high-performance in-memory database that enhances scalability through asynchronous range index updates, offering improved throughput for multi-core systems.
An Extensible Orchestration and Protection Framework for Confidential Cloud Computing
Extensible framework for confidential cloud computing, addressing privacy concerns through hypervisor-level instrumentation, enabling monitoring of enclave-OS interactions and implementation of protected services.
Relational Debugging - Pinpointing Root Causes of Performance Problems
Discover a novel technique for pinpointing root causes of performance issues through relational debugging, offering efficient diagnosis with minimal executions.
EINNET - Optimizing Tensor Programs with Derivation-Based Transformations
Optimizing tensor programs for deep neural networks using derivation-based transformations, outperforming existing optimizers by leveraging a larger search space and automatically creating new operators.
AlpaServe - Statistical Multiplexing with Model Parallelism for Deep Learning Serving
Explore AlpaServe, a novel system for efficient deep learning model serving using statistical multiplexing and model parallelism to handle bursty workloads and improve request processing rates.
Hyrax - Fail-in-Place Server Operation in Cloud Platforms
Innovative datacenter stack enabling servers with failed components to continue hosting VMs, reducing repair needs by 50-60% without impacting performance or scheduling in cloud platforms.
Triangulating Python Performance Issues with Scalene
Discover Scalene: a precise, low-overhead profiler for Python that tracks CPU, memory, and GPU usage. Learn how it helps optimize code and pinpoint performance bottlenecks.
Characterizing Off-path SmartNIC for Accelerating Distributed Systems
Comprehensive study of off-path SmartNIC performance, exploring communication paths and proposing optimization strategies for distributed systems. Demonstrates significant improvements in file system and key-value store applications.
K9db - Privacy-Compliant Storage for Web Applications by Construction
Explore K9db, a MySQL-compatible database designed for automatic GDPR compliance. Learn how it simplifies data privacy management through explicit ownership and sharing semantics.
No Provisioned Concurrency - Fast RDMA-codesigned Remote Fork for Serverless Computing
Explore MITOSIS, a novel OS primitive enabling rapid remote container initialization for serverless platforms, reducing function latency and memory usage while improving state transfer in serverless workflows.
RON - One-Way Circular Shortest Routing to Achieve Efficient and Bounded-Waiting Spinlocks
Innovative spinlock method for multi-core processors, minimizing communication costs and achieving bounded waiting time through one-way circular routing, improving performance in both user and kernel spaces.
NCC - Natural Concurrency Control for Strictly Serializable Datastores by Avoiding the Timestamp-Inversion Pitfall
Innovative concurrency control technique for datastores that leverages natural consistency to achieve minimal costs while guaranteeing strict serializability, avoiding common pitfalls and optimizing performance.