Leveraging Neural Graph Compilers in Machine Learning Research for Edge-Cloud Systems
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866913812058210304 |
|---|---|
| author | Furutanpey, Alireza Walser, Carmen Raith, Philipp Frangoudis, Pantelis A. Dustdar, Schahram |
| author_facet | Furutanpey, Alireza Walser, Carmen Raith, Philipp Frangoudis, Pantelis A. Dustdar, Schahram |
| contents | This work presents a comprehensive evaluation of neural network graph compilers across heterogeneous hardware platforms, addressing the critical gap between theoretical optimization techniques and practical deployment scenarios. We demonstrate how vendor-specific optimizations can invalidate relative performance comparisons between architectural archetypes, with performance advantages sometimes completely reversing after compilation. Our systematic analysis reveals that graph compilers exhibit performance patterns highly dependent on both neural architecture and batch sizes. Through fine-grained block-level experimentation, we establish that vendor-specific compilers can leverage repeated patterns in simple architectures, yielding disproportionate throughput gains as model depth increases. We introduce novel metrics to quantify a compiler's ability to mitigate performance friction as batch size increases. Our methodology bridges the gap between academic research and practical deployment by incorporating compiler effects throughout the research process, providing actionable insights for practitioners navigating complex optimization landscapes across heterogeneous hardware environments. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_20198 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Leveraging Neural Graph Compilers in Machine Learning Research for Edge-Cloud Systems Furutanpey, Alireza Walser, Carmen Raith, Philipp Frangoudis, Pantelis A. Dustdar, Schahram Distributed, Parallel, and Cluster Computing Machine Learning This work presents a comprehensive evaluation of neural network graph compilers across heterogeneous hardware platforms, addressing the critical gap between theoretical optimization techniques and practical deployment scenarios. We demonstrate how vendor-specific optimizations can invalidate relative performance comparisons between architectural archetypes, with performance advantages sometimes completely reversing after compilation. Our systematic analysis reveals that graph compilers exhibit performance patterns highly dependent on both neural architecture and batch sizes. Through fine-grained block-level experimentation, we establish that vendor-specific compilers can leverage repeated patterns in simple architectures, yielding disproportionate throughput gains as model depth increases. We introduce novel metrics to quantify a compiler's ability to mitigate performance friction as batch size increases. Our methodology bridges the gap between academic research and practical deployment by incorporating compiler effects throughout the research process, providing actionable insights for practitioners navigating complex optimization landscapes across heterogeneous hardware environments. |
| title | Leveraging Neural Graph Compilers in Machine Learning Research for Edge-Cloud Systems |
| topic | Distributed, Parallel, and Cluster Computing Machine Learning |
| url | https://arxiv.org/abs/2504.20198 |