Szczegóły publikacji
Opis bibliograficzny
High-performance and programmable attentional graph neural networks with Global tensor formulations / Maciej Besta, Paweł RENC, Robert Gerstenberger, Paolo Sylos Labini, Alexandros Ziogas, Tiancheng Chen, Lukas Gianinazzi, Florian Scheidl, Kalman Szenes, Armon Carigiet, Patrick Iff, Grzegorz Kwasniewski, Raghavendra Kanakagiri, Chio Ge, Sammy Jaeger, Jarosław WĄS, Flavio Vella, Torsten Hoefler // W: SC'23 [Dokument elektroniczny] : proceedings of the international conference for High performance computing, networking, storage and analysis : Denver, CO, USA, November 12–17, 2023. — Wersja do Windows. — Dane tekstowe. — New York : Association for Computing Machinery, 2023. — e-ISBN: 979-8-4007-0109-2. — S. 1–14, [4], art. no. 66. — Wymagania systemowe: Adobe Reader. — Tryb dostępu: https://dl.acm.org/doi/pdf/10.1145/3581784.3607067 [2023-11-15]. — Bibliogr. s. 12–14, Abstr. — Publikacja dostępna online od: 2023-11-11. — P. Renc - dod. afiliacja: Sano Centre for Computational Medicine, Kraków
Autorzy (18)
- Besta Maciej
- AGHRenc Paweł
- Gerstenberger Robert
- Labini Paolo Sylos
- Ziogas Alexandros
- Chen Tiancheng
- Gianinazzi Lukas
- Scheidl Florian
- Szenes Kalman
- Carigiet Armon
- Iff Patrick
- Kwasniewski Grzegorz
- Kanakagiri Raghavendra
- Ge Chio
- Jaeger Sammy
- AGHWąs Jarosław
- Vella Flavio
- Hoefler Torsten
Słowa kluczowe
Dane bibliometryczne
| ID BaDAP | 150154 |
|---|---|
| Data dodania do BaDAP | 2023-11-16 |
| DOI | 10.1145/3581784.3607067 |
| Rok publikacji | 2023 |
| Typ publikacji | materiały konferencyjne (aut.) |
| Otwarty dostęp |  |
| Creative Commons | |
| Wydawca | Association for Computing Machinery (ACM) |
| Konferencja | International conference for High performance computing, networking, storage and analysis |
Abstract
Graph attention models (A-GNNs), a type of Graph Neural Networks (GNNs), have been shown to be more powerful than simpler convolutional GNNs (C-GNNs). However, A-GNNs are more complex to program and difficult to scale. To address this, we develop a novel mathematical formulation, based on tensors that group all the feature vectors, targeting both training and inference of A-GNNs. The formulation enables straightforward adoption of communication-minimizing routines, it fosters optimizations such as vectorization, and it enables seamless integration with established linear algebra DSLs or libraries such as GraphBLAS. Our implementation uses a data redistribution scheme explicitly developed for sparse-dense tensor operations used heavily in GNNs, and fusing optimizations that further minimize memory usage and communication cost. We ensure theoretical asymptotic reductions in communicated data compared to the established message-passing GNN paradigm. Finally, we provide excellent scalability and speedups of even 4–5x over modern libraries such as Deep Graph Library.