Timing of device to device memory transfers


I have been using the profiler to type my GPU programs and memory transfers.
I just tried my first device to device transfer of a 2D array, but the timing of
the transfer does not appear in the profiler log. Should it? If not, how to obtain such
timings? Thanks.


Check the bandwidthTest example in the SDK.


Of course. I did not think of that. Thanks.

The timers use in the bandwidthTest example times only the GPU usage?

I will use this solution.