In the discussion above I'm referring to folks saying that FF "bogs down" or becomes unresponsive in real world usage, something I've never experienced.
But benchmarks are fun too...
this is quantified not only by some arbitrary
number but also through the visibly slower
update speed during the DOM tests.
I ran that and Chrome was 50% faster than FF on my 2018 MBP. Safari was even faster than Chrome by a few %. I would also agree that there was a visible difference in those benchmarks.
My opinion is that while benchmarks are vital, I don't believe that a benchmark such as that necessarily correlates with user experience. There are benchmarks where FF is faster and certainly they can't all be right. (At first glance, the one you linked appears to be a mix of modern UI frameworks, which seems pretty reasonable to me)
A look at a browser's dev tools during my "normal" usage certainly aligns with what I'm saying -- network bandwidth/latency really dominates perceived performance in nearly every use case; blasting out DOM updates as fast as possible via a synthetic benchmark does not resemble my browsing sessions.
That is of course just my subjective opinion. You may have other use cases in which your browser is truly the bottleneck. Or you can just have another set of preferences.
I think it really depends on what you do and where you live. I have access to high-speed internet, have (way too) many tabs open at all times and use various web applications. Someone who's just reading the news and social media behind a DSL line will have a completely different experience.
I rarely encounter loading times that are caused by network traffic. Some servers are slow to reach, but most pages load almost instantaneously. Perhaps that's why things that I see Firefox lose to Chrome at are noticeable to me but not to many others.
The benchmark itself is not an indicator of what the average user will do in their browser, but it does show that Firefox is objectively slower in certain areas. Whether those areas bother you is another question, of course.
Firefox becoming completely unresponsive happens to me more than I like. This is due to a difference in architecture; Firefox spawns a process for every CPU thread and divides the work whereas Chrome spawns a new thread for every page. When a badly written page slows down to a crawl (i.e. leaving Reddit open for more than two minutes), the tabs sharing that process become unresponsive. On multiple occasions I've had to kill tabs that were shown as "loading" because other tabs stopped functioning if they lost the render process lottery. It's not as extreme as people claim and it's certainly not as bad as it was five years ago, but Firefox does have weird freezes and bugs that Chrome seemingly just doesn't have.
On the other hand, I'm convinced part of the reason for slowdowns and problems is that Firefox tries to strip away a lot of the privacy invasions that have become the norm on the modern web. I'll gladly keep taking those slowdowns, but even outside those there's still a lot of work that can be done to bring Firefox on par with Chrome.
I haven't had FF "bog down" (but I'm not really a tab user, either). But on every machine I've put the new FF on, the startup time for it ranges from 2-5 minutes, and the initial load time for web pages is achingly slow (sometimes up to a minute or so).
No other browser I've used performs this poorly. Even the old, pre-reworked, FF performs much better than that.
But on every machine I've put the new FF on, the
startup time for it ranges from 2-5 minutes
What? Is this a typo? Do you mean seconds?
I've never seen that on any machine and again, this is many dozens of computers over the years, including one company where FF was the standard browser for the company enterprise/intranet app so thousands of people were using it all day long.
and the initial load time for web pages is
achingly slow (sometimes up to a minute or so).
It's not a typo. It's also not a particular machine. It's all of my machines.
It's why I stopped using Firefox. It wasn't always like that -- pre-quantum, Firefox was reasonably performant. And post-quantum it was as well (although I never saw the performance gains others reported). I don't remember which release this started happening in, but it was a couple of years ago.
Just to forestall advice -- every time I've mentioned this, people have engaged with me to make sure it's not the usual issues (have I completely erased FF and installed it fresh, am I using extensions, etc.). And many think I'm lying.
But this is my actual experience with Firefox. I was willing to stick with using it on principle despite the fact that post-quantum firefox does a poor job if meeting my browser needs, but I couldn't. It's essentially unusable for me.
Does the issue persist on other networks? I wonder if it's a DNS/proxy/something issue.
I don't know the particulars, but Firefox does some talking over the network at startup. It checks for browser updates, blocklist updates, captive portal detection, etc. I am not certain which if any of these happen at startup, and AFAIK none of these are blocking (but I admittedly have only the vaguest knowledge here)
I've not tried with those machines on a different network. I assume it's related to networking in some way. My network is more complex than most. All of these machines are Linux (Debian). I haven't experienced this on Windows machines at work (although I also haven't noticed that Firefox had become any faster on them).
I wonder what happens if you launch FF with your wifi disabled and/or the cable unplugged?
On MacOS and Windows (I can try this on PopOS as well if you like) FF launches instantly with no complaints if the network is down. Which tells me that FF seems to properly handle situations where things are merely unreachable.
The fact that it's hanging for you implies to me that perhaps FF is getting some kind of malformed response from... something on your network. A firewall, a proxy, I don't know.
I've not tried with those machines on a different
network. I assume it's related to networking in some
way. My network is more complex than most.
I don't know if you're open to feedback on this or not, but it feels disingenuous to state "Firefox is unusably slow" without noting that it appears to be some specific interaction between Firefox and your network. If it is a network issue, it certainly sounds like it is something FF should handle more gracefully/informatively, but that is a different matter from "Firefox unusably slow."
> it feels disingenuous to state "Firefox is unusably slow" without noting that it appears to be some specific interaction between Firefox and your network.
I don't think it's disingenuous because it's only Firefox that has this issue. Every other browser is fine.
I think there's an edge-condition bug in Firefox that I'm triggering that causes this. I've spent quite a while trying to work this out before giving up entirely. However -- since it's really not at all hard to find people who have performance problems with Firefox generally, there is something going on with it. And it's something that (based on interactions with the dev team) Mozilla seems not willing to worry much about.
All that said, I try very hard to be fair to Mozilla and Firefox, in part because I've been a huge backer of Firefox since the very beginning, and still care.
But benchmarks are fun too...
I ran that and Chrome was 50% faster than FF on my 2018 MBP. Safari was even faster than Chrome by a few %. I would also agree that there was a visible difference in those benchmarks.My opinion is that while benchmarks are vital, I don't believe that a benchmark such as that necessarily correlates with user experience. There are benchmarks where FF is faster and certainly they can't all be right. (At first glance, the one you linked appears to be a mix of modern UI frameworks, which seems pretty reasonable to me)
A look at a browser's dev tools during my "normal" usage certainly aligns with what I'm saying -- network bandwidth/latency really dominates perceived performance in nearly every use case; blasting out DOM updates as fast as possible via a synthetic benchmark does not resemble my browsing sessions.
That is of course just my subjective opinion. You may have other use cases in which your browser is truly the bottleneck. Or you can just have another set of preferences.