Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Browse Against the Machine

medium.com

Post reply on HN
Print view View original

131–140 of 142 posts

Re: Browse Against the Machine

#131

Post by RandyRanderson »

1123581321 wrote: post #127
RandyRanderson wrote: post #126

Can someone explain to me why ppl think that splitting the browser application into separate processes will positively affect performance?

1. Separate processes for UI and tab rendering/computation. Same reason iOS is always smooth; in addition to allowing faster access to menus, inspector, extensions etc., it simply feels faster to experience an interface at a constant 60fps. 2. Can pass work off to more CPU cores. I'm not sure of the real world performance gain, but theoretically you could render three heavy tabs and the Firefox UI on a 4 processor CP…

All browsers use multiple threads internally so single process browsers too can use mutliple cores and, in fact, can more efficiently.

Further, most popular modern operating systems handle processes and threads similarly internally.

I'm not sure what you mean by 60 fps. 2d apps typically don't have render loops as games do - they only "composit" graphics typically by updating the UI only if a section needs to be re-drawn.

Re: Browse Against the Machine

#132

Post by jwr »

bzbarsky wrote: post #121
jwr wrote: post #98

Sorry, but until Firefox runs JavaScript code at least on par with Safari, it isn't a good option for some of us. I am the author of a reasonably large web app ( https://partsbox.io/ ), and in real-life performance on a Mac, Chrome is #1, Safari follows right behind it, and Firefox lags far behind. So, Firefox isn't "faster" and doesn't "lag less", quite the opposite.

Just out of curiosity, did Chrome get faster since https://news.ycombinator.com/item?id=13752024 or did Safari get slower? In any case, I'd like to try to figure out what's going on here. Is the problem reproducible with the publicly available demo of partsbox? If so, what are the steps to reproduce? That is, what operations are you measuring?

I don't think either happened. I made a mistake, I intended to say that Safari comes out on top. Sorry about that.

Yes, the public demo is just fine for trying things out. As for measuring, Chrome does seem to have a faster JavaScript engine, as operations like search indexing take shorter than in Safari. However, for interactive use, Safari provides the best experience. I do not know why — perhaps due to the way it updates the displayed DOM? In Chrome long computations in JavaScript run fast, but Safari feels more snappy in real use.

Re: Browse Against the Machine

#133

Post by DavideNL »

cerberusss wrote: post #128
DavideNL wrote: post #8

...meanwhile we still can't use "pinch to zoom" (trackpad) in Firefox, which is available in Safari and Chrome for like 5 years now. Such a basic feature! It's the main reason i personally stopped using Firefox.

Get the extension pinch-to-zoom-firefox-osx https://addons.mozilla.org/en-US/firefox/addon/pinch-to-zoom...

Thanks, but that's kind of a "shitty zoom". What i was looking for is the smooth Safari/Chrome-like zoom, see my previous comment: https://news.ycombinator.com/item?id=14439720

Re: Browse Against the Machine

#134

Post by bzbarsky »

jwr wrote: post #132
bzbarsky wrote: post #121

Earlier quoted context omitted.

Just out of curiosity, did Chrome get faster since https://news.ycombinator.com/item?id=13752024 or did Safari get slower? In any case, I'd like to try to figure out what's going on here. Is the problem reproducible with the publicly available demo of partsbox? If so, what are the steps to reproduce? That is, what operations are you measuring?

I don't think either happened. I made a mistake, I intended to say that Safari comes out on top. Sorry about that. Yes, the public demo is just fine for trying things out. As for measuring, Chrome does seem to have a faster JavaScript engine, as operations like search indexing take shorter than in Safari. However, for interactive use, Safari provides the best experience. I do not know why — perhaps due to the way it…

> Yes, the public demo is just fine for trying things out.

OK. Could you point me to the specific operations on the public demo that are slow in Firefox? I've been trying a few things side by side with Chrome/Safari and not really seeing a perceptible difference, but I don't really know my way around this UI, so it's possible I'm not finding the right things, or not noticing lag that you're noticing... Note that I was using a Firefox nightly; if you're seeing the slowness in a specific Firefox release, which one is it?

Re: Browse Against the Machine

#135

Post by ralphc »

A big use case for me on Chrome is multiple users with their own passwords, cookies etc. At work it's essential for logging in to multiple Salesforce orgs at the same time, and at home I use it to manage multiple gmail accounts. I can't consider using Firefox as my main browser until it has multiple users as easily as Chrome does.

Re: Browse Against the Machine

#136

Post by alttab »

nnethercote wrote: post #124
alttab wrote: post #67

Earlier quoted context omitted.

That's the quote that jumped out to me too. I moved to Chrome about 5-6 years ago, and I'm utterly shocked that they haven't done this yet. I remember this being the reason I switched to Chrome. And they still haven't done it? In 6 years? No wonder I'm still using Chrome...

Adding multi-process support is the single biggest change ever made to Firefox. It was enormous. That's why it took so long.

Took, or taking? It's still not delivered AFAIK.

Re: Browse Against the Machine

#137

Post by jeffbax »

roblabla wrote: post #32
pandatigox wrote: post #28

I'm going add a controversial opinion. I've been using Firefox my whole life but recently I'm finding Safari to be a more appealing option than both Firefox or Chrome. Firefox, and Chrome to an extent, are powerhouses. They have a great community with awesome and really creative extensions. Safari only just started adding extensions and it shows: apart from the usual ad blocker stuff, there's not much out there. But…

Safari being MacOS-only doesn't help its cause very much. It's also so far behind on web standards that some started to call it "the new IE"[0]. Since this, apple started to show some new love to safari in the form of the safari tech preview, but it's still far from where it should be... [0] https://news.ycombinator.com/item?id=12051267

Ahem, https://kangax.github.io/compat-table/es6/

Safari does pretty good I'd say. "The new IE" was pretty overblown.

Safari is my go-to daily driver for speed, OS integration, and battery life. I use Chrome for developing, but maybe I'll give FF another shot.

Re: Browse Against the Machine

#138

Post by sullyj3 »

prodmerc wrote: post #63
yeukhon wrote: post #14

I use Firefox as my main browser at work and at home, but Firefox performance degrades as soon as I have 10 tabs opened. Not FF's entire fault though, a lot of my tabs are JIRA board so there are some stupid Javascript from plugins making the performance more worse. With Chrome, however, I was able to load over 100 tabs easily and I rarely get a crash... So when I need to load lots of tabs, I use Chrome. I still don'…

How do you deal with that many tabs in Chrome? I found it to be a nightmare, the tab buttons are small, you can't even see the titles. Typing a part of a URL you remember in the address bar dumps you to Google search instead of the actual URL. Those two reasons were the only thing keeping me from switching to Chrome when it was fast (apparently it isn't now)...

I generally just use multiple windows. You can click on a tab, then shift click on a different one to select a section that corresponds to some particular task or train of thought, and then drag them all out into a separate window at once.

Re: Browse Against the Machine

#139

Post by cmurf »

abrowne wrote: post #110
cmurf wrote: post #72

I don't have many problems with Firefox desktop, but in particular Firefox on Android (and Chrome to some degree as well) is increasingly a nightmare with news pages. As I'm scrolling and reading, the content jumps around as reflows occur when e.g. an embedded ad or video renders and is squeezed into the content. Instead of rendering the complete page with holes for this embedded content to eventually appear, Firefox…

Two words: Reader View

It helps a ton. But you have to cancel the loading of the page to get that icon. And many of the pages I'm arriving at will do their own reload when I click cancel, so it can be a juggling act to cancel, and in a brief window of opportunity click on the reader view icon before the reload is triggered.

Re: Browse Against the Machine

#140

Post by 1123581321 »

RandyRanderson wrote: post #131
1123581321 wrote: post #127

Earlier quoted context omitted.

1. Separate processes for UI and tab rendering/computation. Same reason iOS is always smooth; in addition to allowing faster access to menus, inspector, extensions etc., it simply feels faster to experience an interface at a constant 60fps. 2. Can pass work off to more CPU cores. I'm not sure of the real world performance gain, but theoretically you could render three heavy tabs and the Firefox UI on a 4 processor CP…

All browsers use multiple threads internally so single process browsers too can use mutliple cores and, in fact, can more efficiently. Further, most popular modern operating systems handle processes and threads similarly internally. I'm not sure what you mean by 60 fps. 2d apps typically don't have render loops as games do - they only "composit" graphics typically by updating the UI only if a section needs to be re-d…

I am sure you are right, but Mozilla says FF used one process, currently uses two (one for UI, one for content) and the final version of multiprocess will use more for web content and extensions. https://developer.mozilla.org/en-US/Firefox/Multiprocess_Fir...

I shouldn't have said 60fps. I just meant low or no interface lag. Regardless of what the system is trying to do internally, Firefox has performance issues in the UI, so hopefully they can address those.

Post reply on HN