It's a fourfold difference (~2200ms versus ~9100ms on three runs of each) on synthetic benchmarks (Sunspider 0.9.1 on an iPhone 4S running iOS 5.1.1, UIWebView was tested through Instapaper 4.2), which I would call significantly worse.
> otherwise nobody would use it.
Performance is not everything: they are convenient especially given iOS's debatable story for switching between applications (especially back and forth between an application and a web browser).
> But lots of apps do use it e.g. Linkedin, Facebook, Reeder, Google Search, iCab Mobile etc and it is not noticeably worse.
It's "not noticeable"[0] for JS-light websites because... well the JIT is for javascript code, running JS slower doesn't matter if there's no JS code.
[0] it is most definitely noticeable, many sites — especially full-page non-mobile ones with ads shit — load significantly faster using Safari than using embedded UIWebView.
Hard numbers would be more interesting.
But lots of apps do use it e.g. Linkedin, Facebook, Reeder, Google Search, iCab Mobile etc and it is not noticeably worse.