All geek questions in one place

Is there a difference between cudaMallocHost () and cudaHostAlloc () without special flags?

At CUDA, I’m interested in learning about the difference between cudaMallocHost()and cudaHostAlloc().

Their respective CVs in the API reference read:

  • cudaMallocHost(): "Allocates memory with a page lock on the host."
  • cudaHostAlloc(): "Allocates memory with a page lock on the host."

And the further description is presented only by additional flags that you can specify for cudaHostAlloc(). Is the former just for backward compatibility or is there a real difference between the two?

PS - Suppose CUDA 6.5 and later.

+4
source share
1 answer

, , cudaHostAlloc, UVA. , API, , , , , , API " ".

+8

Source: https://habr.com/ru/post/1629770/

More articles:


All Articles