Related papers: Venice: Improving Solid-State Drive Parallelism at…
On-disk graph-based vector search (GVS) has become the dominant approach for serving large-scale vector databases at high recall, but prior systems struggle to sustain concurrent search and update throughput on high-dimensional workloads.…
Solid-state drives (SSDs) are well suited for near-data processing (NDP) because they: (1) store large application datasets, and (2) support three NDP paradigms: in-storage processing (ISP), processing using DRAM in the SSD (PuD-SSD), and…
Near-data accelerators (NDAs) that are integrated with main memory have the potential for significant power and performance benefits. Fully realizing these benefits requires the large available memory capacity to be shared between the host…
Checkpointing large amounts of related data concurrently to stable storage is a common I/O pattern of many HPC applications. However, such a pattern frequently leads to I/O bottlenecks that lead to poor scalability and performance. As…
Numerous systems require the capability to switch their operational modes seamlessly without any disruptions. The "Synced Parallel Control Paths" method is an innovative control system architecture designed for seamless mode switching. It…
Intensive communication and synchronization cost for gradients and parameters is the well-known bottleneck of distributed deep learning training. Based on the observations that Synchronous SGD (SSGD) obtains good convergence accuracy while…
With emerging storage-class memory (SCM) nearing commercialization, there is evidence that it will deliver the much-anticipated high density and access latencies within only a few factors of DRAM. Nevertheless, the latency-sensitive nature…
Distributed-memory implementations of numerical optimization algorithm, such as stochastic gradient descent (SGD), require interprocessor communication at every iteration of the algorithm. On modern distributed-memory clusters where…
NAND flash-based Solid State Drives (SSDs), which are widely used from embedded systems to enterprise servers, are enhancing performance by exploiting the parallelism of NAND flash memories. To cope with the performance improvement of SSDs,…
Systolic arrays and shared-L1-memory manycore clusters are commonly used architectural paradigms that offer different trade-offs to accelerate parallel workloads. While the first excel with regular dataflow at the cost of rigid…
Phase-change memory (PCM) devices have multiple banks to serve memory requests in parallel. Unfortunately, if two requests go to the same bank, they have to be served one after another, leading to lower system performance. We observe that a…
By introducing programmability, automated verification, and innovative debugging tools, Software-Defined Networks (SDNs) are poised to meet the increasingly stringent dependability requirements of today's communication networks. However,…
Stream processing applications extract value from raw data through Directed Acyclic Graphs of data analysis tasks. Shared-nothing (SN) parallelism is the de-facto standard to scale stream processing applications. Given an application, SN…
Ultra reliable, low latency vehicle-to-infrastructure (V2I) communications is a key requirement for seamless operation of autonomous vehicles (AVs) in future smart cities. To this end, cellular small base stations (SBSs) with edge computing…
Efficient parallelism is necessary for achieving low-latency, high-throughput inference with large language models (LLMs). Tensor parallelism (TP) is the state-of-the-art method for reducing LLM response latency, however GPU communications…
Convolutional Neural Networks (CNNs) are crucial in various applications, but their deployment on resource-constrained edge devices poses challenges. This study presents the Sum-of-Products (SOP) units for convolution, which utilize…
Compute-in-memory (CiM) is a promising approach to improving the computing speed and energy efficiency in dataintensive applications. Beyond existing CiM techniques of bitwise logic-in-memory operations and dot product operations, this…
Recently, Multipath TCP (MPTCP) has been proposed as an alternative transport approach for datacenter networks. MPTCP provides the ability to split a flow into multiple paths thus providing better performance and resilience to failures.…
Systems that require high-throughput and fault tolerance, such as key-value stores and databases, are looking to persistent memory to combine the performance of in-memory systems with the data-consistent fault-tolerance of nonvolatile…
High main memory latency continues to limit performance of modern high-performance out-of-order cores. While DRAM latency has remained nearly the same over many generations, DRAM bandwidth has grown significantly due to higher frequencies,…