Michael Bauer 

Do you use Project TMs? That should be an extract of your TM based on the content of your file - likely a small-ish subsection of your large-ish TM. Could that help performance?

No, virtually all my work falls into the same type of bucket (software localization), it would be nigh on impossible to unpick that into project TMs that give me the same depth of TM. In any case, once it has properly fired up, the speed of searches during translation is ok.

But that's beside the point. Cancelling a search (and I'm talking about Trados > Translation Memories > select TM > Source Text > Search kind of searches) ought to be a near instantaneous command, irrespective of the size of the file I'm searching through.

No, virtually all my work falls into the same type of bucket (software localization), it would be nigh on impossible to unpick that into project TMs that give me the same depth of TM. In any case, once it has properly fired up, the speed of searches during translation is ok.

But that's beside the point. Cancelling a search (and I'm talking about Trados > Translation Memories > select TM > Source Text > Search kind of searches) ought to be a near instantaneous command, irrespective of the size of the file I'm searching through.

Michael Bauer 

I have no idea how this Trados function (TM Management ⇾ …) looks “under the hood”. If you are willing to try the TM Fusion for Trados app, it does have a search function (“Quick filter”), and the app is very fast. A caveat – when modifying segments I sometimes ended up with corrupted TMs, have not had time yet to find out why. I am just telling you that as a word of caution – try it in a sandbox before you do anything that might cause you grief. There is the option to open a database with read-only access. Not sure what you want to do once you found your segment (copy? edit? delete?).