gpu - CUDA: it is possible for a kernel to return a break to CPU? -
I am using c program
CUDA parallelization, and I was wondering if it is possible to return
to
from to
CPU My program essentially is for the loop and I do several parallel actions within that loop; At the beginning of each iteration, I have to control a variable (just measuring the improvement of the ongoing improvements), which remains at
GPU
My wish is that the variable In order to control the control over
CPU to loop (I have a slight
kernel <<> <1>> ).
I have tried copying that variable back to the CPU and controlling the CPU, but I was afraid, it sloows the overall execution down.
Any advice?
There is currently no way that code running on the CPU-enabled GPU code is hosted on the host CPU While stopping, it is not absolutely clear what you are asking about, I am certain that the answer is simply not because there is no available or interrupted mechanism available in any host party device code.
Comments
Post a Comment