EUVIMEDEUVIMED entdecken
Crossref

Technical Perspective on 'MEMPHIS: Holistic Lineage-based Reuse and Memory Management for Multi-backend ML Systems'

Arun Kumar

ACM SIGMOD Record · 23.04.2026 · 55 (1) · 84-84

DOI: 10.1145/3810900.3810915 ↗

Kurz gefasst

Fachartikel in der Zeitschrift ACM SIGMOD Record, online erschienen am 23.04.2026. Verfasst von Arun Kumar.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

ML deployments in real-world settings are often heterogeneous, spanning local multi-core CPU operations, GPU-accelerated computations, and distributed execution on platforms such as Apache Spark. This heterogeneity arises from practical necessity: a feature engineering step might run locally, a large matrix multiplication offloads to Spark when data exceeds driver memory, and DNN layers execute on GPUs. Or an initial homogeneous setup might evolve over time into a heterogeneous one. This multi-backend reality creates a systems challenge that has received surprisingly little research attention: How to holistically manage computation reuse and memory across backends with fundamentally different execution models, memory hierarchies, and caching primitives?

Zitieren

APA

Kumar, A. (2026). Technical Perspective on 'MEMPHIS: Holistic Lineage-based Reuse and Memory Management for Multi-backend ML Systems'. ACM SIGMOD Record, 55(1), 84-84. https://doi.org/10.1145/3810900.3810915

Vancouver

Kumar A. Technical Perspective on 'MEMPHIS: Holistic Lineage-based Reuse and Memory Management for Multi-backend ML Systems'. ACM SIGMOD Record. 2026;55(1):84-84. doi:10.1145/3810900.3810915

BibTeX
@article{Kumar2026,
  author = {Kumar, Arun},
  title = {Technical Perspective on 'MEMPHIS: Holistic Lineage-based Reuse and Memory Management for Multi-backend ML Systems'},
  journal = {ACM SIGMOD Record},
  year = {2026},
  volume = {55},
  number = {1},
  pages = {84--84},
  doi = {10.1145/3810900.3810915}
}
Kostenlos starten →