Cuda Texture 3D backing cuArray is in Global or Shared memory? Can cuArray be in Shared Memory?

Hi,

I’m using a cuda 3D texture that has a cuArray for it backing 3D LUT. Is the cuArray in global or shared memory. Can it be moved to shared memory and if so how. Looking for the fastest speed for the 3D LUT lookups. Thanks in advance.

Tom

Hi,

It is global memory.

Please noticed that shared memory is kind of stack space with life until a block of a kernel finishes.
And it’s only accessible to the threads within the same block.

To move cuArray into shared memory, you can check this tutorial for information:
https://devblogs.nvidia.com/using-shared-memory-cuda-cc/

Thanks.