Translations:Using GPUs with Slurm/2/fr: Difference between revisions

Jump to navigation Jump to search
no edit summary
No edit summary
No edit summary
Line 11: Line 11:
| 192 || V100L, Cedar || 32|| 192000M || 4  || V100-PCIE-32Go ||  deux GPU par socket CPU; tous les GPU connectés via NVLink
| 192 || V100L, Cedar || 32|| 192000M || 4  || V100-PCIE-32Go ||  deux GPU par socket CPU; tous les GPU connectés via NVLink
|-
|-
| 160 || ''GPU base'', Graham || 32|| 127518M || 2  || P100-PCIE-12Go || un GPU par socket CPU
| 160 || GPU, Graham || 44|| 127518M || 2  || P100-PCIE-12Go || un GPU par socket CPU
|-
|-
| || ''GPU base'', Graham || 28 || 183105M || 8 || V100-PCIE-16Go || quatre GPU socket CPU
| 30 || ''GPU base'', Graham || 28 || 192000M || 4 || Tesla T4-16Go || deux GPU par socket CPU
|-
| 36 || ''GPU base'', Graham || 16 || 196608M || 4 || Tesla T4-16Go || deux GPU par socket CPU
|-
|-
| 15 || K20, Hélios  || 20 || 110000M || 8 || K20-5Go || quatre GPU par socket CPU
| 15 || K20, Hélios  || 20 || 110000M || 8 || K20-5Go || quatre GPU par socket CPU
rsnt_translations
56,430

edits

Navigation menu