rsnt_translations
56,563
edits
No edit summary |
(Created page with "Les versions par défaut des modules suivants ont été mises à jour : * GCC 9.3 => GCC 12.3 * OpenMPI 4.0.3 => OpenMPI 4.1.5 * Intel compilers 2020 => 2023 * Intel MKL 2020 => Flexiblas 3.3.1 (avec MKL 2023 ou BLIS 0.9.0) * CUDA 11 => CUDA 12") |
||
Line 26: | Line 26: | ||
GCC becomes the default compiler, instead of Intel. We compile with Intel only software which have been known to offer better performance using Intel. CUDA becomes an add-on to OpenMPI, rather than the other way around, i.e. CUDA-aware MPI is loaded at run time if CUDA is loaded. This allows to share a lot of MPI libraries across CUDA and non-CUDA branches. | GCC becomes the default compiler, instead of Intel. We compile with Intel only software which have been known to offer better performance using Intel. CUDA becomes an add-on to OpenMPI, rather than the other way around, i.e. CUDA-aware MPI is loaded at run time if CUDA is loaded. This allows to share a lot of MPI libraries across CUDA and non-CUDA branches. | ||
Les versions par défaut des modules suivants ont été mises à jour : | |||
* GCC 9.3 => GCC 12.3 | * GCC 9.3 => GCC 12.3 | ||
* OpenMPI 4.0.3 => OpenMPI 4.1.5 | * OpenMPI 4.0.3 => OpenMPI 4.1.5 | ||
* Intel compilers 2020 => 2023 | * Intel compilers 2020 => 2023 | ||
* Intel MKL 2020 => Flexiblas 3.3.1 ( | * Intel MKL 2020 => Flexiblas 3.3.1 (avec MKL 2023 ou BLIS 0.9.0) | ||
* CUDA 11 => CUDA 12 | * CUDA 11 => CUDA 12 | ||