Related papers: Checkpointing SPAdes for Metagenome Assembly: Tran…
Modern supercomputers are increasingly requiring the presence of accelerators and co-processors. However, it has not been easy to achieve good performance on such heterogeneous clusters. The key challenge has been to ensure good load…
Deployment of real-time ML services on warehouse-scale infrastructures is on the increase. Therefore, decreasing latency and increasing throughput of deep neural network (DNN) inference applications that empower those services have…
Emerging applications of machine learning in numerous areas involve continuous gathering of and learning from streams of data. Real-time incorporation of streaming data into the learned models is essential for improved inference in these…
Many scientific computations need multi-node parallelism for matching up both space (memory) and time (speed) ever-increasing requirements. The use of GPUs as accelerators introduces yet another level of complexity for the programmer and…
Hypergraph partitioning is a recurring NP-hard problem in engineering; its efficient solution at scale hinges on parallelism. This work proposes a GPU-centric algorithm for multi-level hypergraph partitioning aimed at a specific set of…
DeepSeek-V3.2-Exp introduces a sparse attention mechanism that significantly reduces inference latency in long-context scenarios. Although the overall throughput has improved greatly, the Decode-stage of PD disaggregation remains to be a…
The exponential growth of data necessitates distributed storage models, such as peer-to-peer systems and data federations. While distributed storage can reduce costs and increase reliability, the heterogeneity in storage capacity, I/O…
We report optimization work made in support of the development of a realtime Superfacility workflow between DIII-D and NERSC. At DIII-D, the ion properties measured by charge exchange recombination (CER) spectroscopy are required inputs for…
We focus on the problem of checkpointing (or taking a snapshot) in fully replicated eventually consistent distributed databases. In particular, we consider the problem of taking Distributed Transaction-Consistent Snapshots (DTCS). A typical…
In-memory ordered key-value stores are an important building block in modern distributed applications. We present Honeycomb, a hybrid software-hardware system for accelerating read-dominated workloads on ordered key-value stores that…
Recent data stream processing systems (DSPSs) can achieve excellent performance when processing large volumes of data under tight latency constraints. However, they sacrifice support for concurrent state access that eases the burden of…
Cyber-physical systems (CPS) integrate sensing, computing, communication and actuation capabilities to monitor and control operations in the physical environment. A key requirement of such systems is the need to provide predictable…
Traditional public blockchain systems typically had very limited transaction throughput because of the bottleneck of the consensus protocol itself. With recent advances in consensus technology, the performance limit has been greatly lifted,…
InfiniBand is widely used for low-latency, high-throughput cluster computing. Saving the state of the InfiniBand network as part of distributed checkpointing has been a long-standing challenge for researchers. Because of a lack of a…
With growing deployment of Internet of Things (IoT) and machine learning (ML) applications, which need to leverage computation on edge and cloud resources, it is important to develop algorithms and tools to place these distributed…
Multi-tenancy for latency-critical applications leads to re-source interference and unpredictable performance. Core reconfiguration opens up more opportunities for colocation,as it allows the hardware to adjust to the dynamic performance…
As LLMs and foundation models scale, checkpoint/restore has become a critical pattern for training and inference. With 3D parallelism (tensor, pipeline, data), checkpointing involves many processes, each managing numerous tensors of varying…
Analyzing large-scale performance logs from GPU profilers often requires terabytes of memory and hours of runtime, even for basic summaries. These constraints prevent timely insight and hinder the integration of performance analytics into…
In the most popular distributed stream processing frameworks (DSPFs), programs are modeled as a directed acyclic graph. This model allows a DSPF to benefit from the parallelism power of distributed clusters. However, choosing the proper…
Many modern applications require real-time processing of large volumes of high-speed data. Such data processing needs can be modeled as a streaming computation. A streaming computation is specified as a dataflow graph that exposes multiple…