CUDA not starting immediately on Slave - network render

I am a Blender Octane Medical Animator. Windows user. My master computer has 2 3080s, and my slave computer has 2 2080s . All work properly during renders when used independently. Of late, the slave CUDA is not starting until 30 seconds into the render of a frame. As most of my frames are under 20 seconds, this makes my network renders useless. How can I get the CUDA to kick in immediately on the slave? An earlier version of the network render did work this way. All Nvidia Drivers are up to date. Thank you