An official website of the United States government
Communication is a key bottleneck in federated learning where a large number of edge devices collaboratively learn a model under the orchestration of a central server without sharing their own training data. While local SGD has been proposed to reduce the number of FL rounds and become the algorithm of choice for FL, its total communication cost is still prohibitive when each device needs to communicate with the remote server repeatedly for many times over bandwidth-limited networks. In light of both device-to-device (D2D) and device-to-server (D2S) cooperation opportunities in modern communication networks, this paper proposes a new federated optimization algorithm dubbed hybrid local SGD (HL-SGD) in FL settings where devices are grouped into a set of disjoint clusters with high D2D communication bandwidth. HL-SGD subsumes previous proposed algorithms such as local SGD and gossip SGD and enables us to strike the best balance between model accuracy and runtime. We analyze the convergence of HL-SGD in the presence of heterogeneous data for general nonconvex settings. We also perform extensive experiments and show that the use of hybrid model aggregation via D2D and D2S communications in HL-SGD can largely speed up the training time of federated learning.
more »
« less
Understanding Routable PCIe Performance for Composable Infrastructures
Routable PCIe has become the predominant cluster interconnect to build emerging composable infrastructures. Empowered by PCIe non-transparent bridge devices, PCIe transactions can traverse multiple switching domains, enabling a server to elastically integrate a number of remote PCIe devices as local ones. However, it is unclear how to move data or perform communication efficiently over the routable PCIe fabric without understanding its capabilities and limitations. This paper presents the design and implementation of rPCIeBench, a software-hardware co-designed benchmarking framework to systematically characterize the routable PCIe fabric. rPCIeBench provides flexible data communication primitives, exposes end-to-end PCIe transaction observability, and enables reconfigurable experiment deployment. Using rPCIeBench, we first analyze the communication characteristics of a routable PCIe path, quantify its performance tax, and compare it with the local PCIe link. We then use it to dissect in-fabric traffic orchestration behaviors and draw three interesting findings: approximate max-min bandwidth partition, fast end-to-end bandwidth synchronization, and interference-free among orthogonal data paths. Finally, we encode gathered characterization insights as traffic orchestration rules and develop an edge constraints relaxing algorithm to estimate PCIe flow transmission performance over a shared fabric. We validate its accuracy and demonstrate its potential to provide an optimization guide to design efficient flow schedulers.
more »
« less
- Award ID(s):
- 2212192
- PAR ID:
- 10525628
- Publisher / Repository:
- USENIX Association
- Date Published:
- ISBN:
- 978-1-939133-39-7
- Format(s):
- Medium: X
- Location:
- Santa Clara, CA
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Achieving low remote memory access latency remains the primary challenge in realizing memory disaggregation over Ethernet within the datacenters. We present EDM that attempts to overcome this challenge using two key ideas. First, while existing network protocols for remote memory access over the Ethernet, such as TCP/IP and RDMA, are implemented on top of the Ethernet MAC layer, EDM takes a radical approach by implementing the entire network protocol stack for remote memory access within the Physical layer (PHY) of the Ethernet. This overcomes fundamental latency and bandwidth overheads imposed by the MAC layer, especially for small memory messages. Second, EDM implements a centralized, fast, in-network scheduler for memory traffic within the PHY of the Ethernet switch. Inspired by the classic Parallel Iterative Matching (PIM) algorithm, the scheduler dynamically reserves bandwidth between compute and memory nodes by creating virtual circuits in the PHY, thus eliminating queuing delay and layer 2 packet processing delay at the switch for memory traffic, while maintaining high bandwidth utilization. Our FPGA testbed demonstrates that EDM's network fabric incurs a latency of only ~300 ns for remote memory access in an unloaded network, which is an order of magnitude lower than state-of-the-art Ethernet-based solutions such as RoCEv2 and comparable to emerging PCIe-based solutions such as CXL. Larger-scale network simulations indicate that even at high network loads, EDM's average latency remains within 1.3x its unloaded latency.more » « less
-
The confluence of advanced networking (5G/6G) and distributed cloud technologies (edge/fog computing) are rapidly transforming next-generation networks into highly distributed computation platforms, especially suited to host emerging resource-intensive and latency-sensitive services (e.g., smart transportation/city/factory, real-time computer vision, augmented reality). In this paper, we leverage the recently proposed Cloud Network Flow (CNF) modeling and optimization framework to design a novel two-timescale orchestration system for the joint control of communication and computation resources in cloud-integrated networks. The Long-Term Controller solves a properly constructed CNF optimization problem at a longer timescale that determines i) the end-to-end CNF routes (defining data paths and processing locations) for each service chain and ii) the associated allocation of communication and computation resources. The Short-Term Controller uses a local control policy to adjust the allocation of communication and computation resources based on queue state observations at a shorter timescale. Driven by the lack of proper simulation tools, we also develop new ns-3 features that allow modeling and simulation of cloud-integrated networks equipped with both communication and computation resources hosting arbitrary service chains. Finally, we integrate the proposed orchestration system into ns-3 to evaluate and analyze the dynamic orchestration of a set of representative service chains over a hierarchical cloud-integrated network.more » « less
-
Large-scale real-time analytics services continuously collect and analyze data from end-user applications and devices distributed around the globe. Such analytics requires data to be transferred over the wide-area network (WAN) to data centers (DCs) capable of processing the data. Since WAN bandwidth is expensive and scarce, it is beneficial to reduce WAN traffic by partially aggregating the data closer to end-users. We propose aggregation networks for performing aggregation on a geo-distributed edge-cloud infrastructure consisting of edge servers, transit and destination DCs. We identify a rich set of research questions aimed at reducing the traffic costs in an aggregation network. We present an optimization formulation for solving these questions in a principled manner, and use insights from the optimization solutions to propose an efficient, near-optimal practical heuristic. We implement the heuristic in AggNet, built on top of Apache Flink. We evaluate our approach using a geo-distributed deployment on Amazon EC2 as well as a WAN-emulated local testbed. Our evaluation using real-world traces from Twitter and Akamai shows that our approach is able to achieve 47% to 83% reduction in traffic cost over existing baselines without any compromise in timeliness.more » « less
-
As AI systems grow increasingly specialized and complex, managing hardware heterogeneity becomes a pressing challenge. How can we efficiently coordinate and synchronize heterogeneous hardware resources to achieve high utilization? How can we minimize the friction of transitioning between diverse computation phases, reducing costly stalls from initialization, pipeline setup, or drain? Our insight is that a network abstraction at the ISA level naturally unifies heterogeneous resource orchestration and phase transitions. This paper presents a Reconfigurable Stream Network Architecture (RSN), a novel ISA abstraction designed for the DNN domain. RSN models the datapath as a circuit-switched network with stateful functional units as nodes and data streaming on the edges. Programming a computation corresponds to triggering a path. Software is explicitly exposed to the compute and communication latency of each functional unit, enabling precise control over data movement for optimizations such as compute-communication overlap and layer fusion. As nodes in a network naturally differ, the RSN abstraction can efficiently virtualize heterogeneous hardware resources by separating control from the data plane, enabling low instruction-level intervention. We build a proof-of-concept design RSN-XNN on VCK190, a heterogeneous platform with FPGA fabric and AI engines. Compared to the SOTA solution on this platform, it reduces latency by 6.1x and improves throughput by 2.4x–3.2x. Compared to the T4 GPU with the same FP32 performance, it matches latency with only 18% of the memory bandwidth. Compared to the A100 GPU at the same 7nm process node, it achieves 2.1x higher energy efficiency in FP32.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- The DOI is not currently available.
