[Webkit-unassigned] [Bug 120047] Relatively slow performance on emscripten-compiled memops benchmark

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=120047





--- Comment #3 from Filip Pizlo <fpizlo at apple.com>  2013-08-20 11:42:02 PST ---
(In reply to comment #2)
> (In reply to comment #0)
> > Created an attachment (id=209146)
 --> (https://bugs.webkit.org/attachment.cgi?id=209146&action=review) [details] [details]
> > memops benchmark
> > 
> > Attached is the "memops" benchmark from emscripten. The numbers on my 64-bit linux machine are
> > 
> > jsc  5.532 seconds
> > sm   1.551
> > v8   1.488
> > 
> > It is possible to run the benchmark for a longer time period if that is useful, by giving it an argument of 4 or 5 (-- 4 for jsc or v8, just 4 for sm). The results are overall similar.
> 
> Interesting, will look.

OK - I think that something is terribly wrong on Linux, or V8 and SM got a lot faster.  On my 64-bit Mac, I see:

Chrome 28: 3.162 sec
A recent WebKit: 5.382 sec

So, clearly we're slower and we should fix that, but we're not *that* much slower.

Can you check, for sanity, if you see a difference between Chrome 28 (the latest Chrome on the release channel) and the V8 you have?

For posterity I post the HTML-ified version of the benchmark I used to run in browser.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.