The Convoy C-8 has a decent pill and connection to the host and since you have the emitter on a Noctogon there’s no reason you couldn’t safely double stack one side and get closer to 4A. The head will get hot but you won’t have the emitter desoldering and slipping off as with old aluminum stars. If you were planning on running it full blast until it starts to dim then consider a larger host but for some minutes at a time high shouldn’t be a problem. This is why there are now drivers with a turbo mode with step down after a short period, so that you can have the wow mode and stable operation at a more reasonable level. With the light in hand it’s easy to tell when it’s time to throttle back so maybe don’t sit it on a table and leave it on high untended. Be sure the thermal adhesive between the Noctogon and pill is a thin layer(press and twist then check for coverage and clamp in place). A P60 light with it’s smaller head, smaller pill, and poor thermal connection can handle 3A without a Noctogon. FWIW I generally don’t use high for extended periods on any light. As far as run time goes the higher the current the higher the Vf of the emitter and the more voltage sag of the battery so it’s starts to dim sooner. It’s hard to say exactly when that will occur since there are also small voltage drops in the switch, springs, driver, and wires(why we do spring mods and use fat wires) as well as variations in cell capacity. If you were to compare the Discharge curve of that cell (sorry link isn’t working but google discharge curve of NRC 18650b) with the Current vs Forward voltage (scroll down to graphs)curve of your emitter you can see that a tenth of a volt either way makes a big difference. At 4A an XM-L2 is close to 3.5V(maybe less on a Noctogon) and that crosses the discharge curve(between 3A and 5A) somewhere around 1.25 Ahr which figures to about 20 minutes or so. That’s just based on graphs though and again doesn’t figure voltage losses in the light but that’s how you guesstimate it.