Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Swift: A Multi-FPGA Framework for Scaling Up Accelerated Graph Analytics

Oluwole Jaiyeoba, Abdullah T. Mughrabi, Morteza Baradaran, Beenish Gul, Kevin Skadron

TL;DR

Swift is proposed, a novel scale-up graph accelerator framework that processes large graphs by leveraging the flexibility of FPGA custom datapath and memory resources, and optimizes utilization of high-bandwidth 3D memory (HBM).

Abstract

Graph analytics are vital in fields such as social networks, biomedical research, and graph neural networks (GNNs). However, traditional CPUs and GPUs struggle with the memory bottlenecks caused by large graph datasets and their fine-grained memory accesses. While specialized graph accelerators address these challenges, they often support only moderate-sized graphs (under 500 million edges). Our paper proposes Swift, a novel scale-up graph accelerator framework that processes large graphs by leveraging the flexibility of FPGA custom datapath and memory resources, and optimizes utilization of high-bandwidth 3D memory (HBM). Swift supports up to 8 FPGAs in a node. Swift introduces a decoupled, asynchronous model based on the Gather-Apply-Scatter (GAS) scheme. It subgraphs across FPGAs, and each subgraph into intervals based on source vertex IDs. Processing on these intervals is decoupled and executed asynchronously, instead of bulk-synchonous operation, where throughput is limited by the slowest task. This enables simultaneous processing within each multi-FPGA node and optimizes the utilization of communication (PCIe), off-chip (HBM), and on-chip BRAM/URAM resources. Swift demonstrates significant performance improvements compared to prior scalable FPGA-based frameworks, performing 12.8 times better than the ForeGraph. Performance against Gunrock on NVIDIA A40 GPUs is mixed, because NVlink gives the GPU system a nearly 5X bandwidth advantage, but the FPGA system nevertheless achieves 2.6x greater energy efficiency.

Swift: A Multi-FPGA Framework for Scaling Up Accelerated Graph Analytics

TL;DR

Swift is proposed, a novel scale-up graph accelerator framework that processes large graphs by leveraging the flexibility of FPGA custom datapath and memory resources, and optimizes utilization of high-bandwidth 3D memory (HBM).

Abstract

Graph analytics are vital in fields such as social networks, biomedical research, and graph neural networks (GNNs). However, traditional CPUs and GPUs struggle with the memory bottlenecks caused by large graph datasets and their fine-grained memory accesses. While specialized graph accelerators address these challenges, they often support only moderate-sized graphs (under 500 million edges). Our paper proposes Swift, a novel scale-up graph accelerator framework that processes large graphs by leveraging the flexibility of FPGA custom datapath and memory resources, and optimizes utilization of high-bandwidth 3D memory (HBM). Swift supports up to 8 FPGAs in a node. Swift introduces a decoupled, asynchronous model based on the Gather-Apply-Scatter (GAS) scheme. It subgraphs across FPGAs, and each subgraph into intervals based on source vertex IDs. Processing on these intervals is decoupled and executed asynchronously, instead of bulk-synchonous operation, where throughput is limited by the slowest task. This enables simultaneous processing within each multi-FPGA node and optimizes the utilization of communication (PCIe), off-chip (HBM), and on-chip BRAM/URAM resources. Swift demonstrates significant performance improvements compared to prior scalable FPGA-based frameworks, performing 12.8 times better than the ForeGraph. Performance against Gunrock on NVIDIA A40 GPUs is mixed, because NVlink gives the GPU system a nearly 5X bandwidth advantage, but the FPGA system nevertheless achieves 2.6x greater energy efficiency.

Paper Structure

This paper contains 21 sections, 6 figures, 3 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Background and Related Work
  3. Gather-Apply-Scatter (GAS)
  4. ACTS: Near-Memory FPGA Graph Processing
  5. ForeGraph: Scalable FPGA Graph Processing
  6. Swift
  7. Graph Processing Decoupled Pipeline
  8. Understanding the Swift Pipeline
  9. Swift Flow Example
  10. Graph Partitioning and Workload Balancing
  11. Graph Partitioning
  12. Workload Balancing
  13. Performance Evaluation
  14. Experimental Methodology
  15. Graph Algorithms and Datasets
  16. ...and 6 more sections

Figures (6)

  • Figure 1: How Swift (ACTS jaiyeoba_acts_2023 based pipeline) handles Process Edge, Partition-Updates, and Apply Update operations.
  • Figure 2: Swift decoupled multi-FPGA graph execution flow compared to prior art - decoupled operations execute asynchronously on the graph with no bulk synchronization.
  • Figure 3: Swift framework ensures the graph is partitioned into load-balanced intervals distributed across FPGAs and Processing Elements (PEs), and then processed asynchronously.
  • Figure 4: Performance Comparison of Gunrock (GPU) and Swift (FPGA) for PageRank (left), SpMV (middle), and HITS (right) using for 16 iterations (trials) 4 FPGAs/GPUs.
  • Figure 5: Efficiency improvement for Swift over Gunrock --- PR, SPMV, and HITS.
  • ...and 1 more figures