分布式、并行与集群计算

LLM scheduling is critical to serving, yet it remains unclear how well existing designs fit agentic serving--with LLM requests issued by agents instead of humans. This shifts the workload in two ways: (1) agents act only on complete…

Fluid Computing aims to support distributed applications execution across heterogeneous cloud, edge, and device resources, motivating task execution mechanisms that adapt to dynamic and privacy-sensitive environments under runtime…

We initiate the study of deterministic computation in anonymous dynamic networks where each agent broadcasts one bit per round and receives only the number of neighbors broadcasting each bit value. Despite this severe restriction,…

Randomized Kaczmarz is a natural fit for large sparse least-squares and tomographic reconstruction, and adaptive row selection can reduce iteration counts. However, deploying adaptive selection on a shared-memory machine means sampling from…

This work presents ongoing research on the frequency scaling behavior of NVIDIA GPUs when executing ML/AI workloads. Our preliminary findings show that, on lower-performance GPUs, the operating frequency is strongly affected by the recent…

We introduce a distributed computational model referred to as the \emph{uniform port} model. An algorithm operating in this model is defined by means of local automata associated with the ports (a.k.a.\ half-edges) of the input graph. The…

Non-GPU AI accelerators are increasingly adopted as alternatives to general-purpose GPUs for large-model inference, but the real engineering cost of migrating demanding workloads beyond CUDA remains poorly documented. We present a field…

Leadership-class HPC systems are now accelerator-centric, with GPUs providing most floating-point throughput and memory bandwidth. As next-generation systems increasingly integrate accelerators through high-speed memory fabrics and system…

The evolution of compute infrastructure has transformed multi-GPU systems into tightly integrated shared-memory structures. However, current software still mostly treats these coherent interconnects simply as high-speed networks.…

Gradient communication is a primary scaling bottleneck in large language model (LLM) pretraining. Communicating gradients in low-precision formats, such as FP8 and NVFP4, can significantly reduce the communication volume. Existing methods…

Large language models (LLMs) are widely used in intelligent services due to their remarkable capability in generative tasks. Typically, LLM-based services process the inference requests of the users in a centralized data center.…

Container image pulling accounts for the majority of pod startup time in Kubernetes environments. Standard pull downloads the entire image before the container can start, even when the application accesses only a fraction of the image…

The deployment of Mixture-of-Experts (MoE) models on production high-bandwidth superpods, such as NVIDIA's NVL72/576 and Huawei's CloudMatrix384, introduces critical challenges beyond raw interconnect bandwidth. While these systems provide…

Blockchain applications may have preferences over the order in which transactions execute: an automated market maker may use an external feed to price its liquidity, and require that the oracle update incorporating this price execute before…

Interactive debugging is an effective tool for understanding program behavior at the source level, allowing developers to pause execution, navigate the call stack, and inspect runtime state. However, interactive debuggers are designed for…

While CRDTs provide decentralized replication and eventual consistency, Byzantine-resilient deployments require mechanisms for deciding which updates should be trusted and therefore contribute to the reconstructed state. In practice, the…

In this paper, we introduce the \emph{Surplus Parking Gathering Problem} ($\mathcal{SPG}$), a new coordination problem for robots deployed on an infinite grid. The input consists of a set of designated parking nodes, each associated with a…

Matrix optimizers such as Muon are attractive for large-scale training because they can improve convergence and token efficiency over coordinate-wise optimizers. Muon does this by orthogonalizing momentum-smoothed matrix updates with…

Zero-knowledge proof systems rely on a trusted setup phase to generate a Common Reference String (CRS), yet existing approaches are typically static, one-time ceremonies that are inflexible and vulnerable to long-term compromise. Offloading…

AI/ML workloads increasingly run as containers, where a container image must be downloaded to the host before the workload can start. This cold image pull lands on the critical path whenever a training or inference job scales up or a host…