Hello,
I have some 3d memory allocated on the device, using cudaPitchedPtr and cudaMalloc3D.
cudaExtent gridSize = make_cudaExtent(sizeof(float) * 32, 32, 32);
cudaPitchedPtr d_data;
CUDA_SAFE_CALL( cudaMalloc3D(&d_data, gridSize) );
And now I would like to read only a single float value from the memory, so there is no need to transfer all data from the device to the host.
I tried using the
cudaMemcpyFromSymbol
, but that wasn’t successful. Any ideas ?
Thanks.
- Glimberg