I don't think there's enough high quality benchmark information to really make a statement like that, but most importantly, I care about both single-core and multi-thread performance. I don't really have any workloads that only use one thread.
Comparing the M4 with PC CPUs will be hard. Typically when comparing two PC CPUs, to make the comparison more realistic, you'd set some reasonable similar constraints, like using the same memory kits and so on. However, even without considering overclocking, the actual performance of a given CPU can vary massively depending on the thermals, power delivery, memory and so forth. (It can vary by over 50%. I didn't check but you should be able to see this on benchmark charts that allow user submission.)
(However, for what it's worth, I always do at least a bit of mild overclocking personally. Nothing extreme, but what does fit within the power and thermal budget is basically just free performance at the cost of some efficiency, a trade-off I'm happy to make for my main desktop machine.)
Nah that's all pointless trivia. It is dark inside the box. Nobody gives a rip whether the mini is faster because it's got better ram or if it's faster because it's got better arithmetic logic. So you do not have to control things like memory because you don't have a choice anyway.
You don't really seem to understand the point of benchmarks. You're trying to compare the performance between two devices to quantify which one is better at some specific task in some scenario. The tricky part here isn't that people care whether the CPU is better or not, the problem is that on the PC side you can fix the variables between CPUs so that you can just look at the value of individual CPUs, but you can't do that when comparing across PCs and Mac devices. So what do you pick to compare with? There is no correct answer, but there are some answers that are more sensible than others. e.g. you probably don't want to jump massively into another price class.
If money is no object and you just want ridiculous multicore performance it's going to be pretty hard to beat EPYC. Yes, the single core performance is going to be worse; it won't probably be the best even among PC parts, but many use cases gladly take that tradeoff.
Comparing the M4 with PC CPUs will be hard. Typically when comparing two PC CPUs, to make the comparison more realistic, you'd set some reasonable similar constraints, like using the same memory kits and so on. However, even without considering overclocking, the actual performance of a given CPU can vary massively depending on the thermals, power delivery, memory and so forth. (It can vary by over 50%. I didn't check but you should be able to see this on benchmark charts that allow user submission.)
(However, for what it's worth, I always do at least a bit of mild overclocking personally. Nothing extreme, but what does fit within the power and thermal budget is basically just free performance at the cost of some efficiency, a trade-off I'm happy to make for my main desktop machine.)