I cant find the link right now, but there was a nice timings done where Chrome was using less CPU at lower tab counts, but when it increased count, the CPU utilization was considerably higher than FF.
I'll be giving it a fair shake for a few months.
I cant find the link right now, but there was a nice timings done where Chrome was using less CPU at lower tab counts, but when it increased count, the CPU utilization was considerably higher than FF.
I'll be giving it a fair shake for a few months.
Go to: about:telemetry#scalars-tab
Then look at: browser.engagement.max_concurrent_tab_count
(Pretty clever to use telemetry for this, though.)
At a first try Sidebery looks and feels more modern/slick! Might be what I was looking for!
Or more specifically, I'm wondering what change you could be referring to. We've had incremental GC for many years now, which does exactly what you describe. It's true that we keep splitting up more of the uninterruptible pieces into smaller chunks, but I don't recall any major change there recently. (I'm not very good at marketing, am I?)
And according to telemetry, the incremental slices have been working quite well for most people, at least within the last dozen releases or so. We have a budget, and it's rare that we go over it. Not that I fully trust telemetry; if you have counterexamples please file a bug. (I'd love to have a nice set of scenarios that are problematic for the GC. Our telemetry errs strongly on the side of privacy, as it should, so I can't get URLs automatically.)