One is more modern hardware but has smaller heat sinks so it can't run at 100% cpu or it overheats. The other one with older hardware has plenty of cooling, but of course older hardware isn't as fast which is what I'm mainly basing things on since it's the one that's used the most (24/7 since 2019 basically). It has 32gb ram 6 core phenom 2 amd cpu 3.3ghz black edition (on stock clock). It sucks a good deal of power, but it's what I have on hand. I've been using this server for years, and the script hasn't changed to cause the cpu issues. The only changes are after the cpu spiked the first time I added some extra sleeps in possible areas that could spike cpu usage with no change.
Like I said, I can recreate the issue with the official client, it started back up with the cpu problems at the last rs update and it has come and went in the past a couple times or so. The script effected the most by this has barely changed since 2019. I haven't tried other bot clients to see if anyone has detected the issue and figured out a workaround/solution to it. I'm sure it's quite a niche issue that most people don't notice it. If I had more time to learn how a bot client is made, I'd probably be able to figure out what's going on exactly. I know RSA encryption is part of the login protocol, but encrypting a message isn't hard for the cpu and 1000's of messages can be encrypted very quickly. This is more like a wait loop with no sleep in Jagex's code, or they intentionally make it high cpu by performing some useless task like md5 junk data 100k times. It comes and goes with rs updates, so I'm almost positive it's 100% jagex's code side of things. If it's just a bad loop or useless processing, I don't think it would be too hard to skip over the work load or add a small sleep. It could be possible they have a different protocol for login sometimes that requires more work to be done (say the 100k md5 thing, but with the password) and they use that to validate if the password matches what they have pre-processed.
When the cpu issue happens, java is using 100% usage of a single cpu core. Before the same area of the login would take around 5-20% (1-4% overall cpu usage vs 17%).
I suspect in 1-2 rs updates it will be gone again.