Combining Machine Learning and Lifetime-Based Resource Management for Memory Allocation and Beyond.
In: Communications of the ACM, Jg. 67 (2024-04-01), Heft 4, S. 87-96
serialPeriodical
Zugriff:
This article incorporates machine learning into a lifetime-based resource management approach to overcome current limitations of C++ memory allocation. The current state of memory management is first presented with a look into huge pages, long-lived objects, and fragmentation. Then the authors utilized machine learning to predict object lifetime classes for every allocation, then the development of LLAMA—learned lifetime-aware memory allocator—is discussed, and lastly the complete model is evaluated for use.
Titel: |
Combining Machine Learning and Lifetime-Based Resource Management for Memory Allocation and Beyond.
|
---|---|
Autor/in / Beteiligte Person: | Maas, Martin ; Andersen, David G. ; Isard, Michael ; Javanmard, Mohammad Mahdi ; McKinley, Kathryn S. ; Raffel, Colin |
Zeitschrift: | Communications of the ACM, Jg. 67 (2024-04-01), Heft 4, S. 87-96 |
Veröffentlichung: | 2024 |
Medientyp: | serialPeriodical |
ISSN: | 0001-0782 (print) |
DOI: | 10.1145/3611018 |
Schlagwort: |
|
Sonstiges: |
|