You are interpreting the charts correctly. The problem is the charts are ******* trash, almost to the point of outright lies.
Userbenchmark has been heavily messed with to pander to Intel. Never, ever, ever use it.
I didnt know that.
What benchmarks are the correct ones, are they available?
How, specifically, are userbenchmark fiddling their results? I know in the past(coming up 20 years ago now lol), there was Intel Performance Primitives which was a set of libraries that did a lot of mathematical functions, and was great at speeding up certain core inner-loop critical stuff. It turns out that the library was deliberately designed not to work with AMD CPUs, reverting to a much slower version. And when it was cracked/modified to run on AMD it performed as good as equivalent Intel. TLDR; it was synthetically tampered with to make AMD run worse. So if this type of thing is used in a 'benchmark' then I can see that skewing things a lot, and I know it is a trick Intel pulled in the past.
I see the CPU clock cycle numbers there, and I am having a lot of trouble understanding them. It says AMD runs at 4.7 GHz, somewhere it says a lot lower than that, and the overclockers said they could only get 4.3(?) which is lower than the advertised speed? I dont know what is going on. For the Intel, I got one here, and I overclock it to 42x on the multiplier and it will run quite happily at that speed ~4200 on all the cores according to cpu-z(remember I dont know jack **** about serious over clocking, I am just using the tools with the msi and not even the bios). What is going on with AMD clock cycle numbers? What is going to be the actual real clock-speed I can expect?
It seems tho that the 9900KS can reach lets say 5.1 overclocked. People, even amateurs can get this consistently. I certainly would expect this if I buy. And this number seems easy to understand, no fiddling with it.
Let us assume for a moment that these figures are correct and the Intel is running at 5.1 and the AMD at 4.7, all CPU across all cores. This would mean that for the AMD to match the Intel, it would need to do more per cycle than the Intel, or else have something else going on, on the board say, that would compensate for lower clocks. Performance up to 8-core, one would assume, would be better on the Intel, especially in applications which are not parallelised well.
Anyway, I guess if the userbenchmark score is wrong, are the right scores anywhere? How do clock cycle numbers compare, are the CPUs like-for-like on clock cycle, or does one have more effective speed-per-cycle? I know clock cycle is not the be-all as some instructions take more than 1 clock, some can be parallelised, etc.
Thanks guys, I know I am probably annoying some of you by now with my n00by questions.