Use a stopwatch to check how long we are taking on the current ServerCacheQuery batch, if over 2ms then yield for the next frame, or continuously allow processing up to 50 elements per time This (in reality) results in a constant overhead of 4ms until the results are processed rather than a single lag spike. I think its better and can be tuned a bit more in future by just upping the budget By the time you've click the server browser a few seconds have passed and most of the queue is empty anyway