[Webkit-unassigned] [Bug 156147] [JSC] implement async functions proposal

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=156147

--- Comment #94 from Caitlin Potter (:caitp) <caitp at igalia.com> ---
(In reply to comment #92)
> (In reply to comment #90)
> > (In reply to comment #87)
> > > Comment on attachment 279829 [details]
> > > Patch
> > > 
> > > View in context:
> > > https://bugs.webkit.org/attachment.cgi?id=279829&action=review
> > > 
> > > Looks good to me.
> > > Before landing, please ensure that JSBench (you can run it with
> > > run-jsc-benchmark) does not has regression,
> > > since JSBench is good for measuring the code loading (parser) performance.
> > The Parser isn't actually very hot in JSBench because we have
> > the code cache. Octane/code-load (jQuery + closure) are much better
> > indicators of parser performance. Let's run those.
> 
> I usually run this for code-load
> `run-jsc-benchmarks --outer 10 --octane --benchmark "closure|jquery"
> baseline:<pathToVMBaseline> changes:<pathToVMWithChanges>`

well, the results of a mix between comment 91 and comment 92 are this:

Benchmark report for Octane on EchoBeach (MacBookPro11,5).

VMs tested:
"TipOfTree" at /Users/caitp/git/WebKit2/WebKitBuild/Release/DumpRenderTree
"AsyncAwait" at /Users/caitp/git/WebKit/WebKitBuild/Release/DumpRenderTree

Collected 500 samples per benchmark/VM, with 50 VM invocations per benchmark. Emitted a call to
gc() between sample measurements. Used 1 benchmark iteration per VM invocation for warm-up. Used
the jsc-specific preciseTime() function to get microsecond-level timing. Reporting benchmark
execution times with 95% confidence intervals in milliseconds.

                        TipOfTree                 AsyncAwait                                    

closure              0.30106+-0.00079    !     0.30431+-0.00081       ! definitely 1.0108x slower
jquery               2.23843+-0.05744    ?     2.26908+-0.05960       ? might be 1.0137x slower

<geometric>          0.81418+-0.01038    ?     0.82405+-0.01094       ? might be 1.0121x slower

---

Which still isn't great, I guess? It probably gets worse once the runtime flag is removed, too. I'll talk to a colleague about profiling this to see if I can fix it a bit tomorrow/next week.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.webkit.org/pipermail/webkit-unassigned/attachments/20160527/dd5a8001/attachment.html>