one thing you should know before this is, that the millisecond-'timer' aint really that accurate enough to rely on it for showing the exact millisecond you want it to.
so if you just want some more differentiated message timestamp for an about guess of the time differences between 3 or more messages, you could use yourself for a workaround and something like
[[ $+ $asctime(HH:nn:ss) $+ : $+ $right($uptime(mirc,0),3) $+ ]]
as the 'timestamp'.
this wouldnt even show the correct sequence of milliseconds when a second already began, but mirc startet some milliseconds near the end of a second, there could be some 01:872 showing before an 01:065 because mirc started at a milliseconds-counter of 934 second irrelevant.
now as there's no millisecond- with the time-identifiers naturally (besides in $uptime) and if you wanted the 'milliseconds' to start counting at 000, the only way i see is to create that counter yourself, calculated by the milliseconds passed at $uptime and 'recounting' at 000 every time the second changes. and it should only do that when something happens, otherwise mirc will take up half of your machines resources, so you'll get frustrated, stop using mirc, forget to eat and drink, then starve and dehydrate, begin to write unreadable poems, painting senseless pictures, possibly kick the bucket, eat pizza or (even worse) marry someone.
but i wonder if pulling all those messages, notices, actions etc to show up in your own theme-like /echo commands would be worth the while of eliminating any possible 'negative' side-effects and stuff like that afterwards. ever more if you ran a script of some kind which needed to use the geniune mirc functions etc
might seem to be complicated or not, guess you've find out the best way for yourself then.
still have to agree, that its somehow 'pointless' to timestamp in milliseconds. seconds work fine, theres only very seldomly anyone saying two important things in the same second you'd like to know of how fast they were said.
every witless line of code is one witless line of code too much.