- From: Eli Grey <me@eligrey.com>
- Date: Tue, 13 May 2014 13:41:32 -0400
- To: Boris Zbarsky <bzbarsky@mit.edu>
- Cc: whatwg <whatwg@lists.whatwg.org>
On Tue, May 13, 2014 at 1:15 PM, Boris Zbarsky <bzbarsky@mit.edu> wrote: > this proposal seems to assume that the UA itself is using a very > few cores The proposal does not assume anything regarding current system load. If the UA is using every core for some CPU-intensive operation, then navigator.hardwareConcurrency should still be exactly the same value. It represents maximum concurrency potential available, not currently unused concurrency. Once the UA is done with whatever CPU-intensive operations it needs to do, then threadpools previously sized with navigator.hardwareConcurrency will return to nominal performance. Let the OS scheduler do what it does best.
Received on Tuesday, 13 May 2014 17:42:17 UTC