Chrome 79 Is Running Past Firefox 72 Performance On Linux

[GruenSein]

I came here to say something very similar.
Although better performance is always welcome, for day to day usage, such things are imperceptible to the average user. In most cases, your network connection will be the bottleneck. I think once Firefox gets proper GPU acceleration, any performance concerns should be wiped away.

What really matters most is the browser's ability to render sites completely and accurately, while remaining functional. Both Chrome and Firefox accomplish this just fine. Even Edge and Safari could do that, at least at one point. Other features like extensions are also important, and again, Chrome and Firefox have plenty of support.

Extensions are a good point. I used to use Safari - simply because I prefer the native fullscreen behavior on macOS (It opens a new desktop, where the fullscreen video is playing, while the main window remains usable with other tabs, commenting, etc.) - but since Apple has decided to only allow their limited Safari-style extensions, I have switched to Firefox. Other than the trade off extensions vs native fullscreen, I have noticed little to no difference.

[amehaye]

I'm still waiting for a build of Firefox with the V8 JavaScript engine instead of SpiderMonkey. Just to know how much of the speed difference is due to JavaScript.

[aufkrawall]

What UI issues?

https://bugzilla.mozilla.org/show_bug.cgi?id=1606224

https://bugzilla.mozilla.org/show_bug.cgi?id=1588589

[treba]

TL; DR: these benchmarks mean nothing
I don't believe any incremental update like this can make browser much faster.
They are busy with fixing infinite number of bugs, adding some cosmetic improvements, then it's time for the next release, they do some work for stabilizing it.
Anything above that is a miracle and unexpected, and if You retest it every month I warrant You it will be +/- the same for months. Maybe some day they'll rewrite an engine on Rust, but I don't believe it will make it times faster, but let's hope it will be stable enough at least.

Eh? I'd say quite the contrary. First of all rust is not just used for simple rewriting, but for rewrites that would have not been possible / extremely hard before. Take for example the CSS engine, where there have been multiple attempts to rewrite it so it could parallelize more work. In c++ that resulted in too much instability so they never shipped it. With rust they were able to do it and shipped in 57. Suddenly, from one version to the other, Firefox runs cycles around Chrome in that regard (Selenium Style bench).

This kind of stuff happens regularly (just check the 70 release notes, it had big improvements for JS and MacOS) and it's just coincidence that 71 and 72 have been rather quite. There are multi things in the pipeline that will likely switched on by default in the next couple of releases / over the next year, including Webrender, Wayland, partial damage improvements (on Wayland), etc.

Where I agree is that these benchmarks are rather limited for real work usage. Especially many of the JS benchmarks, which AFAIK have little in common with most websites. I'd love to see more benchmarks taking into account things like page load time, responsiveness / smoothness, resource usage etc. Closer to what users actually perceive.

[Flaburgan]

A bit slower to execute but loading way less JS because all the crap is now blocked by default in Firefox <3

[Vistaus]

Which part of Firefox is currently written in Rust? CSS. What were the results of the CSS benchmark again?

I think he meant actual rust, as in "Firefox is rusty/old/etc.".

[Danielsan]

In Kraken, Firefox is faster but then again this is a Mozilla JavaScript benchmark so not too surprising.

While All the other benchmarks are optimized for Chrome, hence you should run real tests with real pages.

[piotr.dobrogost]

Meantime Chrome 79 on Linux freezes so hard that it is unusable for many people – https://support.google.com/chrome/thread/23283471?hl=en

[enigmaxg2]

New version, the same old crap... we know that the donations money isn't going directly to the browser development, but not at what level.

Same performance regressions, same RAM-eating bugs, same battery unfriendlyness, same bizarre CPU usage...same crap as 4 or 5 versions ago.

Mozilla should stop focusing too much on services people may or may not use, and focus on the browser while there are still people using it.

The clock is ticking, we're in route to a web monoculture and Mozilla is still asleep.

[eltomito]

Michael Is the geometric mean for subscribers only? Okay, I'll have to fire up some heavy duty spreadsheet...