English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Relational graph neural networks (RGNNs) are graph neural networks (GNNs)with dedicated structures for modeling the different types of nodes and/oredges in heterogeneous graphs. While RGNNs have been increasingly adopted inmany real-world applications due to their versatility and accuracy, they poseperformance and system design challenges due to their inherent computationpatterns, gap between the programming interface and kernel APIs, and heavyprogramming efforts in optimizing kernels caused by their coupling with datalayout and heterogeneity. To systematically address these challenges, wepropose Pigeon, a novel two-level intermediate representation (IR) and its codegenerator framework, that (a) represents the key properties of the RGNN modelsto bridge the gap between the programming interface and kernel APIs, (b)decouples model semantics, data layout, and operators-specific optimizationfrom each other to reduce programming efforts, (c) expresses and leveragesoptimization opportunities in inter-operator transforms, data layout, andoperator-specific schedules. By building on one general matrix multiply (GEMM)template and a node/edge traversal template, Pigeon achieves up to 7.8xspeed-up in inference and 5.6x speed-up in training compared with thestate-of-the-art public systems in select models, i.e., RGCN, RGAT, HGT, whenrunning heterogeneous graphs provided by Deep Graph Library (DGL) and OpenGraph Benchmark (OGB). Pigeon also triggers fewer out-of-memory (OOM) errors.In addition, we propose linear operator fusion and compact materialization tofurther accelerate the system by up to 2.2x.

PIGEON: Optimizing CUDA Code Generator for End-to-End Training and Inference of Relational Graph Neural Networks