Context
| Sorting algorithms play an essential role in various fields, where their performance is of crucial importance. Genomics is no exception. The performances of three sorting algorithms (Quick, Heap and radix sort) are evaluated against a traditional x86 architecture. One of the key findings is that the PiM architecture not only improves the performance of memory-intensive applications, but also that of speculation-intensive applications. Notably, on a server equipped with 20 PiM UPMEMs, selective sorting on PiM outperforms standard CPU-based selective sorting by an order of magnitude, underlining the substantial gains achievable with PiM technology. |
 |
Code link
Publication
- M. Mognol, D. Lavenier, J. Legriel, Sort Algorithm on Processing-in-Memory, COMPAS 2024, Conférence francophone d’informatique en Parallélisme, Architecture et Système, Nantes, Juillet 2024 [slides]
