Experimentally it has been ~0.8% uncompressed to ~1.4% fully compressed (RMS), full width ~4% compressed with a complicated looking distribution. We measure the energy spread with the "sYAG": https://portal.slac.stanford.edu/sites/ard_public/facet/facilities/Pages/sYAG.aspx
The simulation expectation for the 7mm R56 chicane optics are ~1% RMS fully compressed with <2% full width.
Follow up question: What would be the ratio of correlated energy spread to uncorrelated?
It depends on the compression, fully compressed there should be none of course, but we leave in some for 2-bunch operations.