That's entirely possible. Even the same uarch that's clocked higher can score lower per clock in a memory bandwidth-bound scenario as additional clock cycles aren't doing anything.
It’s possibly also the SLC that was making the difference, like with A15 it has 32MB lmao. (A14, 16MB. M base chips? 8.)
Look, everyone knows it’s perf/GHz but it’s just a colloquial term now, it’s kinda similar to the “nanometer” process naming. Not entirely, because that’s a corporate name that can’t be replaced, whereas we can choose different descriptors for the actual “IPC” values (perf/GHz), but still. Most of us know by now that IPC isn’t really a proper term lol.
this is not massively different from what we have above.
But I agree we should only compare laptop vs laptop chips.
I mean I agree, but for absolute performance reasons for example people get sloppy and will shift from “huh interesting IPC” to “what do you mean, AMD and Intel are like 20% ahead on perf vs [insert mobile laptop core from like QC or speculated future one, and I’m sure they’ll do this when MediaTek/Nvidia have a laptop chip too etc]” and cite charts “8@3 this when it’s practically parity or a smaller margin on mobile. And mobile is most of the client market and a majority of the money as well. You cannot have your cake and eat it too on that, lmao. If you want to talk desktops-only — fine, but that’s less and less of the discourse thankfully and deservedly. It’s just a humongous disservice.
Note: nothingness, I don’t mean *you*. I am using it third person/generically.
It’s a pet peeve because neckbeards from the AMD/Intel caucus love playing fast with this.