CUDA with map <value> key> and atomic operations

Question

As far as I know, I can use C ++ templates in CUDA device code. So, if I use a map to create a dictionary, will the operation of adding new values ​​be atomic?

I want to count the number of occurrences of certain values, i.e. create a codebook with code probabilities.

thank

Macs

Answer 1

You cannot use STL in device code. You can check thrust for similar functionality (in particular, check out the experimental namespace).

, CUDA C ++, , , ( Fermi).

, atomicAdd() , CUDA .

Answer 2

, STL cuda? , . . .