On Mittwoch 18 März 2009, Ahmed Fasih wrote:
Hi, I'm invoking a kernel like using
pycuda.driver.Out and using the
time_kernel=True argument to return the time in seconds. Does this
time include the device-to-host copy time?
No, it doesn't. See driver.py, search for "function_call". If you send me a
patch for time_transfers=True, I'd take it.
Thanks! I'm happy to report a preliminary 165x
speed up over our
existing radar imaging implementation thanks to PyCUDA, developed in 5
Neat--I'd say that was probably worth the effort. :)