I've been looking at this from a whole-perspective and more recently kernel perspective, with what data is publicly available and the performance regression appears to follow Darwin 21.x kernel, present in both iOS17 and Sonoma builds.
Now memory isolation? Plausible, that was my first thought, however memory isolation in Windows-world doesn't come with either the performance regression or battery life loss I've seen upgrading from a fully-patched iOS16 (at the time) or fully patched Ventura (now) to Sonoma and this is hitting both my iPhones SEs (latest) which don't use Intel but rather Apple's own ARM which I suspect with Apple's aggressive security posture they've hardware mitigated memory needs (Spectre/Meltdown) well before they implement memory isolation, but I could be wrong.
I actually think based off some of the symptoms, we're seeing kernel based-performance regression that could be in part due to lesser support for drivers ala AMD video at least in my case in Darwin 21.x as I noted performance regressions only in certain websites, not all of them using Safari. The other culprit? I should've gotten a screenshot but was intent to revert my machine to Ventura (which resolved my performance headaches, and poor battery life); there's a HUGE amount of CPU time going to Apple's cloud sync processes in the background in Sonoma. That's gonna hurt yah on battery life, assuming they're occurring on battery and not just plugged into the AC adapter (again, didn't check, but it's notable in my process history).
In my experience in IT though? This is kernel-based performance regression in any event from the broad symptoms across the OS being degraded and my battery life itself taking a steep hit, be it from poor driver integration with Darwin 21.x vs 20.x, which Apple's going to prioritize Apple Silicon on newer kernels, or, memory isolation which impacts everything (highly probably but not a sole-culprit I feel here), or increased cloud sync activities not occurring on AC adapter. If I had to bet? All the above. Why? Newer Apple Silicon is so power efficient compared to Intel, Apple may feel cloud sync processes can occur on battery now, not just AC as they have traditionally in the iOS world, but may have taken a "hammer" approach where both Intel and Apple Silicon machines do sync processes on battery, not just Apple Silicon; there may be no prerequisite logic implementation here, either by haste or possibly design as notably, poorly performing kernels do spur hardware refresh cycles, which in turn fuel profits. Apple's in no hurry to resolve performance regressions that will spur iPhone or MacBook refreshes... Where I get grumpy is I can't downgrade my iPhone to iOS16 which also appears impacted, albeit to a lesser degree, but it's notably lost around 30% battery life post-iOS17 upgrade and is notably slower with Siri, browsing and directions. Only the last supported iPhones are being permitted iOS16-branch signing going forward; the original iPhone X and iPhone 8 models... Those users on those are already slow enough they need no stimulation to refresh.
