Related papers: Photonic Rails in ML Datacenters
Disaggregating the generation and training stages in RL is widely adopted to scale LLM post-training. There are two critical challenges here. First, the generation stage often becomes a bottleneck due to dynamic workload shifts and severe…
As parallelism becomes critically important in the semiconductor technology, high-performance computing, and cloud applications, parallel network systems will increasingly follow suit. Today, parallelism is an essential architectural…
Hybrid switching - in which a high bandwidth circuit switch (optical or wireless) is used in conjunction with a low bandwidth packet switch - is a promising alternative to interconnect servers in today's large scale data-centers. Circuit…
The problem of optimization of the rolling dynamics model is considered. That providing safe movement at high frequency when interacting with the railway. Moreover, allowing to evaluate the dynamic parameters when designing new and…
Several methods exist today to accelerate Machine Learning(ML) or Deep-Learning(DL) model performance for training and inference. However, modern techniques that rely on various graph and operator parallelism methodologies rely on search…
Training machine learning models in parallel is an increasingly important workload. We accelerate distributed parallel training by designing a communication primitive that uses a programmable switch dataplane to execute a key step of the…
Control parallelism and data parallelism is mostly reasoned and optimized as separate functions. Because of this, workloads that are irregular, fine-grain and dynamic such as dynamic graph processing become very hard to scale. An…
Optical approaches have made great strides towards the goal of high-speed, energy-efficient computing necessary for modern deep learning and AI applications. Read-in and read-out of data, however, limit the overall performance of existing…
The increasing gap between datacenter traffic volume and the capacity of electrical switches has driven the development of reconfigurable network designs utilizing optical circuit switching. Recent advancements, particularly those featuring…
Machine learning training places immense demands on cluster networks, motivating specialized architectures and co-design with parallelization strategies. Recent designs incorporating optical circuit switches (OCSes) are promising, offering…
With the rapid adoption of large language models (LLMs) in recommendation systems, the computational and communication bottlenecks caused by their massive parameter sizes and large data volumes have become increasingly prominent. This paper…
This paper introduces OptiGridML, a machine learning framework for discrete topology optimization in power grids. The task involves selecting substation breaker configurations that maximize cross-region power exports, a problem typically…
The Optimal Power Flow (OPF) problem is integral to the functioning of power systems, aiming to optimize generation dispatch while adhering to technical and operational constraints. These constraints are far from straightforward; they…
Optical circuit-switched networks have emerged as an appealing alternative to electrical fabrics as they can reconfigure the network topology at runtime, reducing communication cost and improving bandwidth utilization. Yet exploiting…
Emerging reconfigurable optical communication technologies allow to enhance datacenter topologies with demand-aware links optimized towards traffic patterns. This paper studies the algorithmic problem of jointly optimizing topology and…
The state-of-the-art topologies of datacenter networks are fixed, based on electrical switching technology, and by now, we understand their throughput and cost well. For the past years, researchers have been developing novel optical…
Network connectedness is indispensable for the normal operation of transmission networks. However, there still remains a lack of efficient constraints that can be directly added to the problem formulation of optimal transmission switching…
When accelerators fail in modern ML datacenters, operators migrate the affected ML training or inference jobs to entirely new racks. This approach, while preserving network performance, is highly inefficient, requiring datacenters to…
Traditional wired data center networks (DCNs) suffer from cabling complexity, lack flexibility, and are limited by the speed of digital switches. In this paper, we alternatively develop a top-down traffic grooming (TG) approach to the…
Thesis Statement: Cost-effective switching and spectrum utilization efficiency have become critical design considerations in optical networks. This dissertation provides in-depth exploration of these important aspects, and proposes…