CUDA: Difference between revisions

Line 115: Line 115:
* Pinned memory can be mapped directly into CUDAspace on ''integrated'' devices or in the presence of some IOMMUs.
* Pinned memory can be mapped directly into CUDAspace on ''integrated'' devices or in the presence of some IOMMUs.
** "Zero (explicit)-copy" interface (can never hide all bus delays)
** "Zero (explicit)-copy" interface (can never hide all bus delays)
* Write-combining memory (configured via [[MTRR|MTRRs]] or [[PAT|PATs]]) avoids PCI snoop requirements and maximizes linear throughput
* Write-combining memory (configured via [[MTRR|MTRRs]] or [[Page Attribute Tables|PATs]]) avoids PCI snoop requirements and maximizes linear throughput
** Subtle side-effects; not to be used glibly or carelessly!
** Subtle side-effects; not to be used glibly or carelessly!