rsnt_translations
56,563
edits
(Created page with "== <code>StdEnv/2020</code> == Cette plus récente version de notre environnement logiciel est celle qui a connu le plus de modifications. Les compilateurs par défaut sont pa...") |
(Created page with "===Amélioration de la performance=== Les binaires générés avec le compilateur Intel supportent automatiquement les jeux d’instructions AVX2 et AVX512. Techniquement, ce...") |
||
Line 31: | Line 31: | ||
{{Command|module load StdEnv/2020}} | {{Command|module load StdEnv/2020}} | ||
=== | ===Amélioration de la performance=== | ||
Les binaires générés avec le compilateur Intel supportent automatiquement les jeux d’instructions AVX2 et AVX512. Techniquement, ce sont des binaires multi-architecture, aussi appelés [https://en.wikipedia.org/wiki/Fat_binary fat binaries]. Ceci signifie que quand vous utilisez une grappe comme Cedar ou Graham qui ont connu plusieurs générations de processeurs, vous n’avez plus besoin de charger manuellement un des modules <tt>arch</tt> si vous utilisez des paquets logiciels générés avec le compilateur Intel. | |||
Many software packages which were previously installed either with GCC or with Intel are now installed at a lower level of the software hierarchy, which makes the same module visible, irrespective of which compiler is loaded. For example, this is the case for many bioinformatics software packages as well as the [[R]] modules, which previously required loading the <code>gcc</code> module. This could be done because we introduced optimizations specific to CPU architectures at a level of the software hierarchy lower than the compiler level. | Many software packages which were previously installed either with GCC or with Intel are now installed at a lower level of the software hierarchy, which makes the same module visible, irrespective of which compiler is loaded. For example, this is the case for many bioinformatics software packages as well as the [[R]] modules, which previously required loading the <code>gcc</code> module. This could be done because we introduced optimizations specific to CPU architectures at a level of the software hierarchy lower than the compiler level. |