While I agree with that, a general CPU benchmark would still be of use to compare different VCPUs. More use than what we have right now, which is only Amazon’s

While I agree with that, a general CPU benchmark would still be of use to compare different VCPUs. More use than what we have right now, which is only Amazon’s marketing claims. As an example, a t4g.nano, a t3.medium, a c6a.large and an r6i.large all have 2 VCPUs. Is the CPU performance exactly the same between all of them? Of course not. We need -a- number to compare these all against. Even if it’s not perfect, it’ll be better than nothing. This is something I’ve wanted to do for ages, just run a benchmark of some sort on a ton of different instances across clouds and publish that info. It really would be useful.