There's a lot of mumbo jumbo in e-drum forums that I'd like to warn against here:

1. There are the latency-cranks who claim that they can detect 2 or 3 milliseconds more or less within a range of 10 milliseconds (e.g. 5 ms vs. 8 ms). and e.g. only find a maximum of 4-5 milliseconds total latency tolerable and with 3 ms more already have a serious or even unbearable problem.
And there are actually people and also commercial suppliers of e-drum systems who present a 1-2ms faster latency of a system as a decisive or at least very important advantage. You can't take that seriously; for commercial providers, it is only used for advertising purposes (They also don't mention the number of milliseconds in their advertising, but say something like “significantly faster”).

This is of course ridiculous, such people (those with non-commercial intentions) are imagining things: they either have no idea what a millisecond is (because it is practically a breath of nothing) or they are making themselves important or want to claim that their perception is different and better than that of "normal" people (and by "normal people" I mean musicians or drummers here). I maintain that such latency cranks will totally fail in a latency blind test (which I offer, e.g. for the realization to stop the swagger, or a bet) (like all those who have already taken the blind test!) (and I bet too! - so far, none of the hardcore cranks, who continue to post their latency mumbo jumbo all over the forums even after such a bet has been offered, have dared to make this bet).

So, don't get worried if you read somewhere what such people are spreading about latency: everything under 10ms total latency is completely unproblematic.

(For comparison: the module sound audio output of the Alesis "Strike" or Alesis "Strata" e-drums was measured by Bonedo: 13ms (which has already left my set limit))


2. And then there are many people in e-drum forums who, in their supposed advisory role in these forums, still want to convince others in 2025 that VST e-drumming is only suitable for home use, because the computer is supposed to crash on stage. What is certain is that these people (I'll call them the computer/tech boomers) will most likely crash their computers (LOL); everyone else who buys a capable system, has a clue themselves or delegates the tech stuff professionally, has no problems with computers on stage and can play to their full potential and enjoy it (including the audience 🙂).
Computers have become so pleasant in terms of size, heat generation and noise (when I think of an M4 MacMini, for example) that there is no difference at all in terms of sensitivity compared to an e-drum module.

"(They also don't mention the number of milliseconds in their advertising, but say something like “significantly faster”)"

➜ By the way, I know a Roland endorser who keeps coming out with “Module is significantly faster than VST” posts. If you then reply to him, yes, it's a whopping 2ms, he doesn't respond anymore LOL. He probably has to do it that way because of his job...

If your system is at around 10ms in total and you feel that there is latency: check your drum samples (your drum sampler) - maybe they are cut sloppily and there is empty space before relevant audio material. (And of course make sure you're not introducing additional latency by audio effects that are not free of latency.)

And this is why is it is a good idea to run a system that provides less than 10ms (let's say 4-8ms): To compensate for the shortcomings of sloppy drum library manufacturers and maybe to compensate for some audio effects that introduce a bit of latency (like the oversampled Flatline2 Clipper by Submission Audio that introduces 1-2ms).

Btw:
Yesterday, pro drummer Randy Black visited my drum studio: He played the VST system there and sound came through PA speakers (so you'll have to add around 5ms sound travel - so the sound response had slightly more than 10ms). I asked him after his performance: "Did you notice any latency" He: "Nope, not at all!" Ask him!
😊