Compilation Report

Not every register is driven by a LUT, and hence placed in its own logic element. You probably have 2,408 LUT+Reg going into Logic Elements, 1,820 LUT-only Logic Elements, and 808 register-only Logic Elements.

I don't remember exactly, but don't think the CIII report estimates what the area would be if you tried packing unrelated registers and LUTs. This naturally can hurt performance though. Try putting the design into a quick LogicLock Region, say 3,500 logic elements in size, and see if it fits. If it doesn't, make it bigger. If it does, try smaller. And keep an eye on the performance.