Live data from Hacker News

Firefox's JIT is getting significantly faster

groups.google.com

251–260 of 286 posts

Re: Firefox's JIT is getting significantly faster

#251

Earlier quoted context omitted.

https://en.wikipedia.org/wiki/Category:Web_browsers_based_on...

They were not asking for a completely random fork, thanks.

My point was, that apparently there are already quite some projects forking ff, so if OP wants to support a mozilla free ff, he has already the chance to do so today.

Re: Firefox's JIT is getting significantly faster

#252
post #180
post #173

Earlier quoted context omitted.

In this link about a similar (or same?) issue it sounds like filters are done on the CPU in Firefox unless you enable a flag but it may be outdated: https://bugzilla.mozilla.org/show_bug.cgi?id=925025 > Safari and Chrome both do blur on the GPU Firefox does it in software unless WebRender is turned on. You could try turning on the gfx.webrender.all pref and that should improve things.

webrender is on by default for Windows 10, but Mozilla is still struggling with mac and Linux drivers, so it's still off by default there. See https://wiki.mozilla.org/Platform/GFX/WebRender_Where To check if you're using webrender, check about:support. Check the graphics>compositing section. (Edited to correct first paragraph.)

It is changing, though.

I have an Intel video card (UHD 630), and Webrender (which I had to forcibly enable) is working very well. Video acceleration, too. Even with a 4K screen.

Re: Firefox's JIT is getting significantly faster

#253

Earlier quoted context omitted.

A 10-15% speed bump in CPU performance is certainly nice but in most cases it's not going to be something that users are consciously aware of. Web page performance varies a lot due to network performance, new versions of websites being deployed, different ads being served, and so on. This noise obscures things enough that it probably won't be easy to attribute a change in performance to the browser if you're not look…

Agreed for most cases. However, it matters in the context of canvas/WebGL games.

Firefox used to be a lot faster, and smoother, at rendering than Chrome when I was developing early versions of my canvas library (around the mid 2010s). But then Chrome caught up and for the past couple of years has been considerably faster[1].

One thing that really annoys me about Firefox on MacOS is that they disable keyboard focus of links by default. Which is not good for accessibility - it took me days to work out why my accessibility testing demos were failing[2].

[1] - I use this demo for my goto stress test: https://scrawl-v8.rikweb.org.uk/demo/canvas-006.html

[2] - https://discourse.mozilla.org/t/keyboard-focus-of-links-on-m...

Re: Firefox's JIT is getting significantly faster

#254

It's sad how these valuable optimizations are unknown to the average person. When I hear "the browser" discussed in almost any parlance, the implication is Chrome. It's rare to even hear someone say "Chrome," as it's the defacto choice for the non-mobile web. Convenience breeds ignorance.

Even for the mobile on Android.

Re: Firefox's JIT is getting significantly faster

#255
post #12

Nice! There's still cool cpp stuff going on in Firefox. This is one example. https://searchfox.org/mozilla-central/source/js/src/jit/Warp...

That’s a 3300 line cpp file, care to cite a specific example?

The entire file contains WarpBuilder, which is behind the JIT speedup that this thread is about.

> The SpiderMonkey (JS) team has been working on a significant update to > our JITs called WarpBuilder (or just Warp) [0,1].

The whole component was the cool stuff I was referring too. I linked it mainly to show that it's C++ instead of, say, Rust, not to highlight any component of it.

Re: Firefox's JIT is getting significantly faster

#256
post #181

Earlier quoted context omitted.

I've literally seen people commenting that they don't use firefox because chrome seems faster. (...and also, how on earth did you get it to use that much memory? I'm sitting over 500 tabs, albeit mostly unloaded, and 6 extensions, on a system with only 6GB installed)

Just curious (not being snarky): what is the advantage of kepeing 500 tabs open? How does it work? I rarely have more than 10 open, and I've been using tabs ever since tabs on browsers were invented. (I was using browsers before tabs existed) I feel I might have missed some trend here.

I think for over 500 tabs, one should just use bookmarks. That said, I do have more than 100 tabs open, properly managed through stacked tabs and so forth.

Re: Firefox's JIT is getting significantly faster

#257
post #148

The question for firefox was never been is it fast. The question for firefox was always “why its using 10GB RAM?”

Good job! My FF with 10+ extensions only use 493mb with 4 tabs opened. What am I doing wrong?

Not enough tabs. :D

Re: Firefox's JIT is getting significantly faster

#258

Is there any way to directly donate to firefox (and not other mozilla related things)?

no

And even if there were, you could be subsidizing Mozilla, because they can simply give Firefox less now that you're supporting it

Re: Firefox's JIT is getting significantly faster

#259

Earlier quoted context omitted.

If they want to actually improve user experience they can always include uBlock Origin by default - its permissive license should allow this just fine.

I wonder if blocking Google Analytics by default would affect their default search engine deal with Google...

There is a move toward 1st party proxys as a result of GDPR and similar ruling. Google has already published a framwork or something in that direction. It is not mandatory yet, but when the web moves in that direction there will be no blocking of domains any more. (Technical) Users won't even know that Analytics tracks them.

Re: Firefox's JIT is getting significantly faster

#260

Earlier quoted context omitted.

Do you think that focusing on more unsexy things would have helped? It seems to me that it's fundamentally a marketing problem. Many people are not even aware that Firefox exists, despite it being basically as good as Chrome, and in some ways considerably better.

ultimately if there value , word of mouth works pretty well . Firefox does not have a brand visibility problem , no amount of marketing will convince users if the browser is not significantly better. If the value is marginal convenience trumps. Firefox did not gain market share orginally because they did better marketing than IE. They became big because they built a better product. Chrome became because of the same r…

Firefox did gain over IE because major websites pointed IE users to it.

And of course Chrome became dominant the same way.

Word of google.com main page > word of mouth :(

Post reply on HN