With less efficiency, hence less output, and I want it for everyday use so throw is less of a factor than flood. My EDC right now is the new TrustFire R5-A3 (with the 0.15A low mode) which throws a lot more than I need for EDC. More flood would be just as welcomed as a higher output with the same heat and same draw from batteries as what I have now.
Also I personally am not assuming that XM-L has less throw than an XP-G for the same wattage, since the XP-G in a small reflector doesn't have great throw anyways, and XM-L is more effecient especially around 1.5A where it's at half its rated power and still really high on the efficiency curve. The XP-G at 1.5A is at the high end of its power handling, and well into where the efficiency starts to drop off.
I'm comparing my R5-A3 with fresh "flames" right now on high (1.7A), to my KD C8 w/ fresh flames on medium (1.1A), with a ceiling bounce... and the XM-L is *still* a tad brighter.
However... This does ignore one important aspect that I haven't seen discussed anywhere yet... The psycho-visual effect of pulse width modulation upon subjective brightness. Like when you're at a movie, and each frame blends into the other frames without flickering, even though the shutter is only open for a fraction of 1/24th of a second. :) This *does* increase the subjective average brightness!!!
This is something I was going to talk to SPAMBOT about, since he seemed interested in efficiency, and to see/search if any research has been done on this, and then post a well formed write up about it for you guys. And how to accurately measure it as close as possible to the subjective brightness experienced by the average subject. :) But for now, it seemed important to bring up here, since I was just comparing a current & PWM mode with each other.
Personally I think the subjective brightness (and therefore battery life) of any current regulated modes can be optimized with pulse width modulation being done at a frequency that makes the most use of this particular psycho-visual phenomenon. A driver that maximises the balance between peak current and pulse width, essentially. For starters, dropping the peak current drawn to the most efficient spot for a particular LED, and widening the pulse width to compensate for the average amps drawn, should cause the light to be brighter with the same power used.
This should also have some effect on the way the batteries work too, since we're not using super conductive batteries the current will be drawn differently, and hence they have *some* inductance & capacitance, in addition to also having a base resonance and harmonics in which standing waves can be built up (but making use of that is way more advanced than we'll probably ever see in a flashlight). Mainly the effect would be in reducing the peak current draw during each pulse. :)
Third and final thing that I would like to explore along those lines is the effect that different frequencies of PWM have on the LED itself, from kHz to gHz.
But I'm not ready to get into any of that yet, before I get scientific ammo. I just thought, in the name of science, that I should explain this for people that haven't thought about it yet, that it probably did bias my comparison. If it biased it anywhere near the 63% difference in current pulled (to which the XP-G had the advantage), I couldn't tell ya. Yet. :)
Anyways... :P back on topic. An XM-L is quite a bit more efficient than an XP-G @ 1.5A. More output & less heat. (plus the heat is spread over a larger die, so it should be less prone to heat-induced efficiency reduction for the same overall heat output) Combined with a reflector designed for it, and that alone with current driver designs is still enough to make me want to upgrade.