CUDA.jl - When to synchronize

[maleadt]

@ww1g11

copyto!(y_cpu, y)

Copying to CPU memory automatically synchronizes.

@ww1g11

    kernel1!(backend)(x; ndrange=length(x))
    
    KernelAbstractions.synchronize(backend) # sync1 

    kernel2!(backend)(y, x; ndrange=length(x))

Kernel execution is ordered on the task-local stream, so there’s no need to synchronize in between.