Javascript Performance Thread

[trieste]

Just want to say kudos and amazing to the devs working on Ion. Even on the Aurora build and a Samsung 830 SSD, the improvement in snappiness is still unmistakable, which is really mind-boggling.

[RyanVM]

Thanks for the update, Hannes.

[Zlip792]

Improved performance on SS after long time as well. Great going devs...

[Timvde]

2) What's the deal with JaegerMonkey
It will get removed . Yeah I'm really happy with that. There are some reasons for this. Many optimizations in JaegerMonkey are hacked together, because there is no decent way to do. IonMonkey has a much easier/better design in that regard. Also JaegerMonkey needs to recompile whenever types changes ... That's one of the reasons of our bad performance on SS. Also it looses it's information when we compile IonMonkey, resulting in the performance differences of 10% on some V8 benchmarks. ETA is when the replacement baseline compiler is done and at least as fast as JaegerMonkey.

Will it be as fast as JM+TI? I thought it wasn't meant to be that fast, just to have super fast compile time and no need to recompile, with IM ready to kick in when the workload becomes too heavy.

Improved performance on SS after long time as well. Great going devs...

They finally fixed the performance hit caused by source code compression: bug 836515. It's not really a performance gain, just undoing their big regression

[iwod]

Will JS ever be fast enough? I finally got to a point thinking it never will be in a Multi Tab environment. There are FAR too many little snippet of JS running in the background on every page and in Multiple Tab this just stalled the UI.

[Grantius]

Will JS ever be fast enough? I finally got to a point thinking it never will be in a Multi Tab environment. There are FAR too many little snippet of JS running in the background on every page and in Multiple Tab this just stalled the UI.

Its very fast at the moment - even with 100 tabs it rarely gets as bogged down as it used to.

[Omega X]

That's nothing to do with JS. More like the single threaded DOM.

[helkaluin]

Will JS ever be fast enough? I finally got to a point thinking it never will be in a Multi Tab environment. There are FAR too many little snippet of JS running in the background on every page and in Multiple Tab this just stalled the UI.

That would depend on scheduling instead of raw throughput.

Case in point: With the BFS scheduler, the Linux kernel can run loads of CPU starving processes in the background without sacrificing foreground latency.

I don't know how is scheduling implemented within Firefox, but I'm sure with the ongoing work with Off Main Thread This and Off Main Thread That, it's only going to improve.

[jandem]

Will it be as fast as JM+TI? I thought it wasn't meant to be that fast, just to have super fast compile time and no need to recompile, with IM ready to kick in when the workload becomes too heavy.

Yes that's correct. The baseline compiler alone will be slower than JM+TI, because its design is much simpler (no register allocation, no function inlining etc). Advantages are that it will be much easier to maintain, have faster compile times, collect type information, reduced memory usage and share code with IonMonkey.

Once the baseline compiler is in better shape, there will be some major TI changes, for instance bug 804676. That should also help benchmark performance and memory usage.

[iamhere]

Arewefastyet.com doesn't update charts for MacPro 32/64 bit. Seems the issue is on arewefastyet.com code, as I can reproduce with Chrome.

[bksening]

Err... I'm seeing AWFY updates for Mac Pro for both 32 and 64 bit where it goes up to Feb. 10 in the dates:

Mac Pro 32-bit http://www.arewefastyet.com/#machine=11

Mac Pro 64-bit http://www.arewefastyet.com/#machine=12

[iamhere]

Right, what I meant is that the same charts were being shown for both machines. But on visiting again now, I can't see the issue anymore.

[Josa]

Baseline Compiler now on AWFY!

[Grantius]

Baseline Compiler now on AWFY!

How do we enable it on nightly?

[greg86]

Baseline Compiler isn't already included in mozilla-central, its only in the ionmonkey branch.
Links on awfy to the (BC+Ion+TI) revisions & changelogs are broken