Does loading preloading N values into FMAC X1 buffer use x*N CPU cycles or is it offloaded to the FMAC?

[KHarb.1]

I need to reload the entire X1 buffer between FMAC operations. When the filter coefficient buffer is reloaded, does the FMAC do that in the background when a reload command is given or does the loading process run on the CPU?