A Taxonomy for Classification and Comparison of Dataflows for GNN Accelerators

Garg, Raveesh; Qin, Eric; Muñoz-Martínez, Francisco; Guirado, Robert; Jain, Akshay; Abadal, Sergi; Abellán, José Luis; Acacio, Manuel E.; Alarcón, Eduard; Rajamanickam, Sivasankaran; Krishna, Tushar

doi:10.2172/1817326

Cited by 9 publications

(7 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For (b) representation, the neighbors of a particular vertex are stored back-to-back because they are highly sparse. (b) is often used as the graph representation [5,6]. In this study, we used (b) to describe the adjacency matrix comprising an edge array Fig.…”

Section: Gnnsmentioning

confidence: 99%

Anomaly detection analysis based on correlation of features in graph neural network

Ko,

Praca,

Choi

2023

Multimed Tools Appl

View full text Add to dashboard Cite

Various studies have been conducted to detect network anomalies. However, because anomaly signals are determined by the pattern characteristics using the dataset, the real-time detection problem continues. Even if there is a signal with an attack sign among the constantly transmitted and received signals, the attack cannot be blocked in advance. Moreover, it appears in many places in a distributed denial-of-service (DDoS) attack, so the real-time defense must be the best option. Therefore, it is necessary first to discover the characteristics and elements regarded as abnormal signals to discover anomalies in real time. Finally, by analyzing the correlation between network data and features, extracting the elements of the anomaly, and analyzing the behavior of the extracted elements in detail, we aim to increase the accuracy of the anomaly. In this study, we used Coburg intrusion detection and KDDCup datasets and analyzed the correlation of elements in the dataset using a graph neural network. The calculated accuracy values of the anomaly detection were 94.5% and 98.85%.

show abstract

Section: Gnnsmentioning

confidence: 99%

Anomaly detection analysis based on correlation of features in graph neural network

Ko,

Praca,

Choi

2023

Multimed Tools Appl

View full text Add to dashboard Cite

show abstract

“…For instance, HyGCN [12] explored both intra/inter-vertex parallelisms to separately handle the irregularity in the aggregation phase and reusability in the combination phase. Later, aiming to boost the overall hardware utilization, AWB-GCN [5] proposed to balance the workload during runtime with an auto-tuning algorithm and to increase the data locality by regionally clustering the non-zero values (i.e., connected neighbors) within the adjacency matrices; EnGN [45] proposed a ring-edge-reduce dataflow to handle graphs with arbitrary dimensions and increase the accelerator's scalability to large graphs; and GRIP [46] employed finegrained vertex-tiling to reduce the weight bandwidth requirements; In parallel, to reduce the human efforts in designing GNN accelerators and democratize the process, pioneering works have attempted to characterize the design space of dataflows and micro-architectures for GNN accelerators [13], and developed an automated framework to generate GNN accelerators [14]. Nevertheless, existing automated frameworks for GNNs still have limited support to various GNN structures and thus suffer from low hardware utilization and achievable efficiency on certain tasks.…”

Section: Related Workmentioning

confidence: 99%

“…For example, HyGCN [12] proposes hybrid execution patterns of GNNs for leveraging their intra-vertex and inter-vertex parallelisms to handle the irregularity in the aggregation phase and reusability in the combination phase, respectively; Later, AWB-GCN [5] identifies the workload imbalance problem in the aggregation phase, and proposes auto-tuning workload balancing techniques, achieving an average speedup of 7.4× over HyGCN. On the development tool level, pioneering works have attempted to characterize the design space of dataflows and micro-architectures for GNN accelerators [13], and develop an automated framework to generate GNN accelerators [14].…”

Section: Introductionmentioning

confidence: 99%

G-CoS: GNN-Accelerator Co-Search Towards Both Better Accuracy and Efficiency

Zhang¹,

You²,

Fu³

et al. 2021

Preprint

View full text Add to dashboard Cite

Graph Neural Networks (GNNs) have emerged as the state-of-the-art (SOTA) method for graph-based learning tasks. However, it still remains prohibitively challenging to inference GNNs over large graph datasets, limiting their application to large-scale real-world tasks. While end-to-end jointly optimizing GNNs and their accelerators is promising in boosting GNNs' inference efficiency and expediting the design process, it is still underexplored due to the vast and distinct design spaces of GNNs and their accelerators. In this work, we propose G-CoS, a GNN and accelerator co-search framework that can automatically search for matched GNN structures and accelerators to maximize both task accuracy and acceleration efficiency. Specifically, G-CoS integrates two major enabling components: (1) a generic GNN accelerator search space which is applicable to various GNN structures and (2) a one-shot GNN and accelerator co-search algorithm that enables simultaneous and efficient search for optimal GNN structures and their matched accelerators. To the best of our knowledge, G-CoS is the first co-search framework for GNNs and their accelerators. Extensive experiments and ablation studies show that the GNNs and accelerators generated by G-CoS consistently outperform SOTA GNNs and GNN accelerators in terms of both task accuracy and hardware efficiency, while only requiring a few hours for the end-to-end generation of the best matched GNNs and their accelerators.

show abstract

“…Later, AWB-GCN [13] identifies the workload imbalance problem in the aggregation phase, the non-zero values (i.e., connected neighbors) in adjacency matrices are regionally clustered, and proposes autotuning workload balancing techniques to alleviate the runtime imbalance. Another trend is to summarize the design space of the dataflow and microarchitecture optimization in GCN accelerators [12], and provide automated framework to generate suitable hardware for the given GCN applications [24], [50]. For example, G-CoS [50] develops the first co-search framework that can automatically search for the matched GNN structures and accelerators to maximize both task accuracy and acceleration efficiency.…”

Section: Related Workmentioning

confidence: 99%

“…Why GCN Inference Is Inefficient. There exists a fundamental dilemma associated with GCN inference ac-celeration: To accelerate GCN inference, the irregularity of GCNs' adjacency matrices need to be reduced, which can inevitably degrade the inference accuracy; On the other hand, maintaining GCNs' irregularity and thus their excellent accuracy can lead to extremely high hardware costs of GCN inference as demonstrated in recent works [42], [13], [25], [24], [20], [7], [12], [47]; both limiting their more extensive applications.…”

Section: Gcod: Motivation and Overviewmentioning

confidence: 99%

GCoD: Graph Convolutional Network Acceleration via Dedicated Algorithm and Accelerator Co-Design

You¹,

Geng²,

Zhang³

et al. 2021

Preprint

View full text Add to dashboard Cite

Graph Convolutional Networks (GCNs) have emerged as the state-of-the-art graph learning model. However, it can be notoriously challenging to inference GCNs over large graph datasets, limiting their application to large real-world graphs and hindering the exploration of deeper and more sophisticated GCN graphs. This is because real-world graphs can be extremely large and sparse. Furthermore, the node degree of GCNs tends to follow the power-law distribution and therefore have highly irregular adjacency matrices, resulting in prohibitive inefficiencies in both data processing and movement and thus substantially limiting the achievable GCN acceleration efficiency. To this end, this paper proposes a GCN algorithm and accelerator Co-Design framework dubbed GCoD which can largely alleviate the aforementioned GCN irregularity and boost GCNs' inference efficiency. Specifically, on the algorithm level, GCoD integrates a split and conquer GCN training strategy that polarizes the graphs to be either denser or sparser in local neighborhoods without compromising the model accuracy, resulting in graph adjacency matrices that (mostly) have merely two levels of workload and enjoys largely enhanced regularity and thus ease of acceleration. On the hardware level, we further develop a dedicated twopronged accelerator with a separated engine to process each of the aforementioned denser and sparser workloads, further boosting the overall utilization and acceleration efficiency. Extensive experiments and ablation studies validate that our GCoD consistently reduces the number of off-chip accesses, leading to speedups 15286×, 294×, 7.8×, and 2.5× as compared to CPUs, GPUs, and prior-art GCN accelerators including HyGCN and AWB-GCN, respectively, while maintaining or even improving the task accuracy. Additionally, we visualize GCoD trained graph adjacency matrices for a better understanding of its advantages.

show abstract

A Taxonomy for Classification and Comparison of Dataflows for GNN Accelerators

Cited by 9 publications

References 36 publications

Anomaly detection analysis based on correlation of features in graph neural network

Anomaly detection analysis based on correlation of features in graph neural network

G-CoS: GNN-Accelerator Co-Search Towards Both Better Accuracy and Efficiency

GCoD: Graph Convolutional Network Acceleration via Dedicated Algorithm and Accelerator Co-Design

Contact Info

Product

Resources

About