htop on our vms and clusters, because it's in all the repos, it's fast, it's configurable by a deployable config file, it's very clearly laid out and it does everything I need. I definitely would not call it bloated in any way.

My config includes network and i/o traffic stats, and details cpu load type - this in particular makes iowait very easy to spot when finding out why something's racking up big sysloads. Plus, it looks very impressive on a machine with 80 cores...

My brain can't parse top's output very well for anything other than looking for the highest cpu process.

But - ymmv. Everyone has a preference and we have lots of choice, it doesn't make one thing better or worse than another.