It varies. Flip Chip for instance allows you to choose what percentage of blue and red you want in your cobs. If they were even close to Samsung, Vero29C, Citizen etc. in terms of efficiency they would be worth taking a look at.
@Randomblame does make a very good point though. It prompted me to dig up some old data and look at it again. I've arranged it into a summary chart.
View attachment 4227308
Rows 3-5 - spectral analysis
Row 7 - Raw PPFD readings
Row 8 - Raw readings with Mcree correction factor.
Rows 10, 13, 16 - Relative intensity: % * raw reading / 100
Rows 11, 14, 17 - (10, 13, 16) with Mcree correction factor.
It would have been nice if 4000K 80CRI had been included in the mix, but we can predict with some degree of accuracy the values based on the available data. This wouldn't be true of 5000+ because the spectral curves become very different. If we extrapolate 4000K 80CRI from the data the Blue/Green/Red %s would be around 16.5/46/37.5. PPFD reading would be about 936 and with Mcree correction it wouldn't be much different than 3000K 80CRI or 3500K 80CRI.
CRI- 80CRI provides about 10% red (600-700nm) boost over 70CRI. 90CRI provides about 10% red boost over 80CRI. CRI has a minor impact on blue output. This is all at the expense of the 500-600nm regions. 3000K 70CRI Has the highest par readings by far. Even when a correction factor based on the Mcree curve is applied 3000K 70 CRI still has the highest score due to all the light in the 500-600 region. The sample with the highest R + B is 3000K 90CRI.
K- Blue levels drop about 40% going from 3000K to 2700K. Blue levels increase by about 15% from 3000 to 3500K. 2700K has a slight increase over 3000K in red. 3000K 80CRI has around 8% more red than 3500K 80CRI.
From the samples I think 3000K 70CRI would be a good dedicated veg spectrum, also 4000K 70 or 80CRI. Any of them will get the job done but it wouldn't be intuitive to use 2700K 90CRI for instance, but it might make sense to use 3000K 70CRI full cycle. Regarding flowering and yield, I have no idea which is best. On paper 3000K 70CRI should provide the best yield (YPF) once the phosphor losses are factored in. In practice nobody seems to have shown one to be better grow after grow, Just different and conflicting reports.
20 par watts per foot from most any light source is going to do a really good/similar job. I prefer 3000K 90CRI but more because it seems like the ideal spectrum, rather than because of some guarantee it will work better. I don't care to have UV in my grow space and don't see the point in anything beyond 700nm except perhaps for lights out. If a lamp came with end of day functionality that would be a nice selling point. DIY there are various options for that and UV bulbs if UV is desired.
Why one of the cool kids hasn't built a blurple lamp and qualified it against white after all these years of people DIYing, I don't know. I guess there's a practicality to the design and build process that precludes building things one might not want to use after it's built, or would cost more than necessary to do the same job. Still, it's entirely possible to DIY a high efficiency 450+630+660 lamp and quantify the output to a high degree of accuracy.
