The mars efficacy might be ffair, up to 2.3, but i call bullshit on sf, and so does most people i know who have made real tests of this sort.
The hlg diablo has twice the diodes, a more efficient driver, and handpicking the best samsung diodes; they score 2.81 in verifiable tests. If sf really had highestt flux, lowest voltage bin and half the diodes they would clock in soomewhere around 2.5 -2.6 in system efficiency just by extrapolating from hlg. And i doubt they actually have those best bin diodes.
And in any case: both fixtures lack the spread that 6 points of light would give.
4x sf1000 would be a better buy.