Optimizing the Impact of Upgrading Computer Equipment
Résumé
Upgrading hardware platforms and infrastructure , like personal computer, servers, or supercomputers suffers from a tension when it comes to carbon neutrality. Increasing performances are an incentive for fast renewal, while manufacturing environmental costs urge us to extend as much as possible the lifetime of devices. Thus, carbon neutrality seems to be difficult to reach. In this article, we try to answer quantitatively to the question "what is the optimal point for which we may tend to the carbon neutrality of infrastructures"? Our approach is based on a model of the greenhouse gas emissions, the primary energy and the monetary cost of two parameters that vary country by country: the use of a microprocessor, and its fabrication. We have investigated three different renewal scenarios: the maintain of the workload capacity, the maintain of the number of processor cores, and the maintain of the number of microprocessors. Finally, we apply our model to two concrete CPUs. The most efficient solution (i.e with the lowest impacts) is to replace the entire batch of CPUs with the newer batch as soon as possible. This optimal-replacement strategy depends on the specifications of CPUs and allows to determine the optimal scenario.
Origine | Fichiers produits par l'(les) auteur(s) |
---|