Tobias Benz <[email protected]> writes: > I would like to point you to a question I posted on stackoverflow: > > http://stackoverflow.com/questions/12259010/measuring-time-using-pycuda-driver-event-gives-wrong-results > > Ian Ozsvald and me are not sure whether this is expected behaviour or a > bug. > > It would be great if you could have a look at it!
Does anything change when you synchronize on the start event, too? Nvidia's docs don't really define the behavior for this case. http://developer.download.nvidia.com/compute/cuda/4_2/rel/toolkit/docs/online/group__CUDART__EVENT_ga324d5ce3fbf46899b15e5e42ff9cfa5.html#ga324d5ce3fbf46899b15e5e42ff9cfa5 sort of sounds like your usage is fine... Andreas _______________________________________________ PyCUDA mailing list [email protected] http://lists.tiker.net/listinfo/pycuda
