Some time ago I did a Uniquefire UF-T20 mod with a dedomed XP-G2 R5 (3C-tint) led on a Sinkpad at 2.95A (measured tailcap). I wanted to make a similar second pill for this light (using a 'C8 pill' obtained from Fasttech) with a dedomed XP-E2 R3 (1D-tint) led on a Sinkpad at 2.4A (measured tailcap), so I could make a direct comparison. (There is no gain in driving the XP-E2 any harder than at 2.4A)
Let's do the math first.
Using the famous emitter test results done by match, the emitter output for the XP-G2 at 2.95A should be about 785lm and the emitter output for the XP-E2 at 2.4A should be about 480lm. So (with XP-G2@2.95A and XP-E2@2.4A) XP-G2 puts out 1.64 times as much light. But the die surface of the XP-E2 is 2.2 times smaller than the XP-G2:
so the surface brightness (and thus the expected throw) of the XP-E2 at 2.4A is 2.2/1.64=1.34 times brighter than the XP-G2 at 2.95A. This difference is not huge, just slightly seen by eye, and you have a much smaller hotspot. In other words: there is no sound reason to go XP-E2. Even more, the XP-G2 can be driven quite a bit harder than the used 2.95A, that should be a much better upgrade .
But now abandon theory.
I did the XP-E2 pill anyway (dedomed XP-E2 R3 1D-tint on Sinkpad reflowed onto C8-pill with 7x7135 Nanjg-driver, measured 2.4A at tail):
Uniquefire UF-T20 beamshots compared, left the ddXP-G2 pill, right the ddXP-E2 pill (I made the left photo a while ago):
Measurements:
XP-G2-pill XP-E2-pill
lumen estimate zoom out: 490 lumen 260 lumen
lumen estimate zoom in: 332 lumen 179 lumen
throw at 1 meter, 1 minute: 106 klux 135 klux
throw at 1 meter, settled: 101 klux 128 klux
Looking at these numbers, when going from the XP-G2-pill to the XP-E2 pill, the difference in lumen output is 0.54 (the math said 0.60), the difference in throw is 1.27 (the math said 1.34). So the XP-E2 pill that I made (as compared to the XP-G2-pill) is performing a little bit less than the theory predicts, but I am actually quite happy that it is so close to the expected numbers (it is nice when theory and practice match).
A sad observation is that you loose a lot of light between the emitter and the outside world in a aspheric, even when zoomed out it looks already like around 40%, and zoomed in 60%. EDIT : sorry, I forgot that these emitters are dedomed, that gives at least 20% loss, so an aspheric's efficiency is not that bad after all .
My overall conclusion is that not just in theory, but also in the real world, when the goal is throw, there is not much gain in using dedomed XP-E2 over dedomed XP-G2. (unless you are especially looking for that extreme narrow hotspot, which sure is fun to play with )