Have not used EVE-NG, but (many) years ago, first time I used GNS3 it had a feature to analyze a running IOS that would dramatically reduce its CPU usage. Don't know if GNS3 still has this feature or whether EVE-NG has anything similar.
I do see on EVE-NG site's home page there's mention of the CPU Watchdog feature and release notes mentions a fix daemon capdog/cpulimit, but couldn't find additional details.
I too, like Jim (@Ramblin Tech) had thought to mention CML as an alternative, but as you didn't mention what license variant of EVE-NG you were running, didn't know whether CML's subscription fee would be an impediment. (If CML still offers, they had provided a temporary web lab version that's free.)
Also, I've never loaded up the CSR1000v node, and wanted to try doing so, which I did a short time ago.
My personal PC, on which I run CML is Windows 10, 16 GB RAM with a 4 GHz Quad core AMD Ryzen 3 2300X, likely much less powerful than Jim's system.
Further, I run CML under VM Workstation v5.
When I booted up a CSR1000v node, it took several minutes (pegging my one core). Once booted up, CPU usage fell to under 2%, but memory footprint was about 45% of the 8 GB I've allocated to the VM.
When I attempted to boot up another CSR100v, it failed for lack of hardware resources, memory I presume.
I stopped the first instance, and attempted to boot the two CSR100v nodes, concurrently.
Surprisingly, this succeeded, using about 90% of my lab's memory, CPU fell to about 2%, but I kept getting CML controller pop-ups that its launch was being paused because of over commitment. Did get a (apparently) working command link on both CSR1000v nodes, though.
