To be honest, i find the gram per watt not so useful because i consider it arbitrary. Biased on a way, on a fews points.
By example : a (good) Scrogger than use a 600HPS/sqm versus a (good) Sogger than use 600HPS/sqm will always end in favor of the Scrogger. But the most productive of the both will be ever the sogger, he will have a yield/year difficult to challenge for the first guy. A good and an effective training in SCROG need time, and generally is made under 24/0.
The difference become dramatic if the sogger only use seeds to SOG in a perpetual and single 12/12 space.
Another thing than is pernicious, it's the real watts in question.
On the paper i'm building a 1Kw CFL /sqm space. But this type of tool say than it's in fact a 560W/sqm (4x140w and not the 4x250w than the marketing say).
I'm not op on LED technologies and new HPS, but i think than the drivers/ballast don't eat really what is writed on bulbs/strips/leds. I will be very interrested to see someone digging the subject with his tools, a watt meter is pretty cheap and easy to use.
Add to it, than generally only the light is take in count. Not the active AC/exhaust/intake/pumps/ventilators etc ... which will be a pain to make a ratio with it honestly ^^
On breeding (well it's the subject here no ? ^^), the most usefull ratios i've found to make a yielder was the trimming loss ratio and the grammer/time ^^ That's very personnal off course, but sometimes you have to consider than you have to use the right compass to find the North you want lol
Just a little rationnal share to push you on your very interresting projects ;o)