How to efficiently merge two vectors in alternating way using CUDA thrust

Accelerated Computing CUDA CUDA Programming and Performance
1

My question is how to efficiently merge two vectors alternating indexes in a batch using CUDA thrust.

Toy example:
For two device vectors, vecA=[1, 2, 3, 4, 5, 6, 7, 8, 9] and vecB=[1, 1, 1, 2, 2, 2, 3, 3, 3].
Given batch size len=3, combine vecA and vecB into a new vector vecC.
vecC=[{1, 2, 3}, {1, 1, 1}, {4, 5, 6}, {2, 2, 2}, {7, 8, 9}, {3, 3, 3}].
{} is only for explanation, vecC is a 1-dimensional device vector.

I’m new to CUDA programming, could you provide the code so that I can test the performance?

2

You could use two calls to thrust:scatter .

1 Like
3
1 Like
4

I didn’t wait for an answer here, so go there and ask a question.
I will close this question.
Thanks!

5

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.