After enabling background WebContent process suspension we're measuring up to 1% regression in PLT-warm runs. I think this is due to calling prepareToSuspend on the WebContent process.
<rdar://problem/107494626>