On 06/21/2015 05:38 AM, Nicholas Nethercote wrote:
On Sat, Jun 20, 2015 at 1:36 PM, Aaron Klotz <[email protected]> wrote:
I did some measurements last year at the 2014 JS workweek and it turns out
that the cost of converting the omnijar's JS source to jschars during
parsing is equivalent to the benefits we gain from StartupCache. Assuming
that nothing has changed in SM with respect to the time required to do the
conversion, if we eliminate that conversion step, we can eliminate
XDR-encoded JS from the StartupCache.
Would that mean XDR encoding could be removed altogether? If so, I
will gleefully volunteer to do the removal :)
No because, we will probably want to do that for cold start-up of web-pages
& application, and I do not expect website/app developers to encode anything
in UTF-16.
Also, note that the start-up cache that we currently have does not use a
sparse-bytecode cache, which means that we could potentially optimize it by
only storing functions which are used during the start-up time of the
browser. (making this cache smaller, on B2G, only 33% of the function were
executed during the start-up)
--
Nicolas B. Pierron
_______________________________________________
dev-platform mailing list
[email protected]
https://lists.mozilla.org/listinfo/dev-platform
