memcpy on device is there a way to use STL memcpy on tne device?

Accelerated Computing CUDA CUDA Programming and Performance
1

Hello,

when i try

memcpy(vs, v, sizeof(float) * 100);

the compiler doesn’t complain but it fails at runtime. Is there a way to make this work or do I have to write a loop?

Kind Regards

2

Is this inside a kernel? You can’t use host functions inside kernels from what I recall, but I have been away from CUDA for a bit. If you want to use something like std::vector<> inside a kernel you have to make it a pointer, and allocate the std::vector<> on the device. This may or may not work, it is just something I thought of while replying.

std::vector<float>* vec;

cudaMalloc((void**) &vec, sizeof(std::vector<float>) * size);
3

I seem to remember we actually do support memcpy in kernels (although not STL), but I wouldn’t recommend using it in general.

4

Take a look at Thrust Library: Google Code Archive - Long-term storage for Google Code Project Hosting.

Don’t know if is there what you want, but it makes memory management very easy.