rsnt_translations
56,430
edits
CUDA tutorial/fr: Difference between revisions
Jump to navigation
Jump to search
Created page with "== Stratégies usuelles de programmation == * Constant memory also resides in DRAM - much slower access than shared memory ** MAIS, elle est cachée !!! ** highly efficient ac..."
(Created page with "== Bande passante == * Always keep CUDA bandwidth limitations in mind when changing your code * Know the theoretical peak bandwidth of the various data links * Count bytes rea...") |
(Created page with "== Stratégies usuelles de programmation == * Constant memory also resides in DRAM - much slower access than shared memory ** MAIS, elle est cachée !!! ** highly efficient ac...") |
||
| Line 180: | Line 180: | ||
* Utilize the various memory spaces depending on the situation: global, shared, constant | * Utilize the various memory spaces depending on the situation: global, shared, constant | ||
== | == Stratégies usuelles de programmation == | ||
* Constant memory also resides in DRAM - much slower access than shared memory | * Constant memory also resides in DRAM - much slower access than shared memory | ||
** | ** MAIS, elle est cachée !!! | ||
** highly efficient access for read-only, broadcast | ** highly efficient access for read-only, broadcast | ||
* Carefully divide data acording to access patterns: | * Carefully divide data acording to access patterns: | ||
** read-only: constant memory ( | ** read-only: constant memory (très rapide si dans la cache) | ||
** read/write | ** read/write dans le bloc : mémoire partagée (très rapide) | ||
** read/write | ** read/write dans le fil : registres (très rapide) | ||
** read/write | ** read/write en entrèe/résultats : mémoire globale (très lente) | ||