L2T v2.0 w/P2D body: How's your experience?

I was just thinking of getting an L2T v2.0 or L1T v2.0 w/a P2D body & tailcap. 140 lumens or 90 lumens on 'high,' 15 lumens on 'low,' according to Fenix. Sounds like the perect EDC light. I just want 2 modes, no strobes, sos... and a clicky.

I'm assuming that using a primary 3.0 volt CR123, I would get virtually the same output levels and runtimes as an L2T v2.0 using two 1.5 volt AAs. But would this not actually be the case? I've read one post where this combo seemed to drop out of regulation in like 20 minutes and to virtually no output in about an hour. Don't know if the battery used there was a primary CR123 or an RCR123. A couple of other posters said they use this combo, and like it, but no one has commented on runtime or regulation on a CR123.

Comparing Fenix's specs on the L2D CE and the P2DCE I noted that while the output levels are the same, the 2 x AA version seems to offer almost twice the runtime of the 1 x CR123 version (assuming the heads, emitters, and circuitry are the same, which I think they are). So, at worst, I would think that my L2T/P2D combo will give me L2T level brightness on 'high' and 'low' with the shorter runtimes of the L1T (1.5 hours high/15 hours low)? This wouldn't be bad at all, actually.

Can someone more knowledgeable explain why (if) performance would be different on a primary CR123a vs. 2 x AA? I could understand that it might actually be somewhat brighter compared to using two 1.2 v AAs, but if the light works in regulation on 2 1.5 v AAs, why wouldn't it work the same way on 1 3.0 v CR123? I know there are a few others out there using this combo. Have you experienced this very short runtime/no regulation on 'high' when using a primary CR123?

Maybe I'm imagining a problem where there isn't one, but I don't want to do this and then find it doesn't actually work.

(I know the simple answer is get a Lumapower LM33 instead, but I don't think it would be nearly as bright on 'high' (<90 lumens) or as low on 'low' (??? lumens) and the runtime on 'low' would be much shorter - Lumapower says LM33 runs in low for 4 hours. Let me know if I'm wrong, you could save me some money!)

I'm using a L1Tv2.0 with a P2D body and tailcap, and the reverse clicky from the L1T, as my EDC. Haven't used primaries, though. Instead I'm using AW RCR123a's. Runtime is not shortened on high, as far as I can tell. Haven't specifically tested, but I do get around one hour on high.

On RCR's, low isn't really low until the cell is almost depleted, so runtime on low will be significially shorter than on primaries.

My very subjective judgement of output is that I do get about 140 lumens on high (difficult to differentiate from my P2D Q5 180 lumens, at least in the hotspot). Low on RCR's might be as high as 100 lumens on a fresh cell, untill voltage drops into the regulated region.