Similar to what MrGman is doing by posting lumen values at 30/60/120/180 second intervals, to show the drop in lumens due to loss of efficiency caused by heat sag, I've done something similar using a ceiling bounce test.
I ran the tests twice to confirm the results.
Lights: JetBeam M1X, Sunlite 16wFP, EagleTac M2X, ElektroLumens P7 LongThrow.
Light Meter: Extech 403125
Here are the results:
As you can clearly see, Sunlite rises above the rest in terms of heat removal from the LED. EagleTac comes in 2nd. M1X in 3rd. P7 LongThrow is 4th.
Since these measurements rely mainly upon the 100% value, I'll explain how I gathered that initial "Start" number.
With the meter on and in place, and in a totally dark room, I had the stopwatch in hand and the light in the other. I clicked the light 'On' and set it down, then watched the meter numbers rise until they maxed out. I then started the stopwatch. I recorded down the max number for the 'Start" value, then at 30 seconds, I recorded what the meter read, and etc, etc, for the rest of the times.
You can still see that the M2X is reading the highest value for output, and the 16wFP shows to have the best heat removal.
For the record, the P7 LongThrow was using new Duracell primaries. The M2X and M1X were using AW 2200mAh protected 18650's, and the 16wFP was using Sunlites own battery, a protected 5200mAh.
Edit:
I'll also throw in these beam profile shots that I captured during the bounce tests.....
They are about 6ft away from the ceiling.
M1X and P7 LongThrow:
M2X and 16wFP: