Nightly Firefox Win 64 bit Discussion

[watdis]

FF is perfectly fine when it's using 2GB of RAM, 300 tabs. With Azure and electrolysis coming, I'm not sure if FF will be unresponsive ever, maybe only in extreme cases. With HTML5 and lots of media on internet in right now and in the future, I don't think powerusers will be fine with 32bit build of FF.

IMO.

[mikedl]

FF is perfectly fine when it's using 2GB of RAM, 300 tabs. With Azure and electrolysis coming, I'm not sure if FF will be unresponsive ever, maybe only in extreme cases. With HTML5 and lots of media on internet in right now and in the future, I don't think powerusers will be fine with 32bit build of FF.

IMO.

I agree with your opinion, particularly the latter portion of your above quote. I run many 64 bit applications that require far more than 3GB of Ram depending on the task (AutoCAD Architecture being but on example).

[Fabfire]

Plugins are 3rd party software that Mozilla doesn't have control over.

What I said is that precisely because those applications "are 3rd party software that Mozilla doesn't have control over" that they will appear as the Win64 Firefox is released and get a real user base. As small as they can be on the beginning, or labeled as "power users" - those are the fellows that will make it happen. That's how markets work. If there is a demand for something, offer will follow. The sheer number of Firefox users gives Mozilla leverage to do this.

The browser should not fail to load a plugin that the user already has installed just because the architecture is wrong.

I totally disagree with that. The browser should work with plugins that are built for the architecture that it was built for - 64 bit, in this case. If that is a "sine qua non" premise, then we are doomed - because we'll not have a Firefox x64 in a foreseeable time. Maybe after Google Chrome releases a x64 version (which probably won't work with 32 bit plugins, either).

Flash "Square" is obscenely out of date and has many known security flaws.

No kidding... And what version of Flash Player has no known security flaws? Obscenely out of date? Maybe, if Firefox x64 already existed, it wouldn't be so.

Using it is highly discouraged.

If " known security flaws in Flash Player" is reason to "highly discourage" its use, then all Flash Player versions use should be "highly discouraged"...

This goes back to the first point. Because plugin venders are reluctant to support 64bit architectures, browsers need to support 32bit plugins.

No. Absolutely not. Plugin developers are taking their time and avoiding development costs because browser developers are reluctant (some irracional fear and lame excuses maybe?) to support 64 bit architectures - even when they already have done most of the work, like Mozilla - and when there are a large number of people asking for it. That's the real situation.

So you want Mozilla to release a product, that isn't ready for prime-time, in order to compete with another product that is equally unready for actual use? That logic doesn't make any sense.

My logic makes perfect sense. For some reason, you are not following. I don't know where did you read that I wrote that I "want Mozilla to release a product, that isn't ready for prime-time". You are putting words in my mouth, and then say my logic doesn't make sense? Well, let's go on...

Mozilla should not rush to release a product before it's ready. That's akin to shooting your self in the foot. Plus, there's no rush to have a native 64bit version on Windows, as the 32bit version runs perfectly fine.

Well, let's try it again. I wrote:

Releasing the Win64 Firefox would bring another (not small) advantage - the competition is IE x64, which is - well, weak. Or are we going to wait the release of Chrome x64 to start worrying about it?

What did I meant? That releasing Firefox x64 - meaning as soon as possible, after throwing the lame excuses out ( I NEVER wrote release it as it is) - they would drive a hard blow on their competitors. And I'll add this: Mozilla needs to think seriously on getting back to the road of innovation. This was what made Firefox the fantastic success that it still is. Trying to copy the competitors (for some reason, Google Chrome comes to mind) is never an option for first place. A strategy for innovation is.

[mikedl]

Alright, Fabfire ... I gotta do it (sorry no revelations or additional content - my apologies) but ...

+1,000,000 again!

Thanks.

BTW, Bluefang, your extension (Status4Evar) works wonderfully on 64 bit Windows Fx.

[Bluefang]

This goes back to the first point. Because plugin venders are reluctant to support 64bit architectures, browsers need to support 32bit plugins.

I disagree. There are many applications that are 64-bit. I have a few right now. Xfire chat, FireFox Nightly, Java, EVGA Precision, CoreTemp, Crysis, audio and video drivers, WinRAR, etc. There is no end to the number of already available applications and drivers available in 64-bit. Mozilla in general is being quite lazy since Safari has been 64bit for a while, and IE9 does have a 64bit option. Even Google has a 64bit version in the works.

I think you completely missed the point of what I said. I never said anything about Mozilla producing a 64bit browser. I just said that it needs to support 32bit plugins. And Mozilla agrees, which is why there hasn't been an official Win64 release. Arguing about priorities is a completely different matter.

FF is perfectly fine when it's using 2GB of RAM, 300 tabs. With Azure and electrolysis coming, I'm not sure if FF will be unresponsive ever, maybe only in extreme cases. With HTML5 and lots of media on internet in right now and in the future, I don't think powerusers will be fine with 32bit build of FF.

That's not a particularily convincing argument.

* Azure is just a rendering API that fronts OpenGL, DirectX, D2D, Cairo, or any other rendering backed they desire. Firefox already uses all of these rendering backends on the various platforms. Azure is just going to change how Firefox uses them.

* Electrolysis probably won't see that drastic of a resource overhead. It's not going to be process-per-tab (à la Chrome).

* HTML5. What? So sites are going to start using <audio> and <video> instead of Flash. That's a good thing, and will probably improve resource usage.

And, in this context, "power user" is largely becoming a euphemism for those who must run the latest and greatest just because it is shiny and new. A true power user would know that 64bit is useful in certain specific applications and use-cases. A browser is not [currently] one of them.

BTW, Bluefang, your extension (Status4Evar) works wonderfully on 64 bit Windows Fx.

Why wouldn't it? It doesn't have any binary components. If it wasn't working correctly, it would be due to a flaw in the Win64 version of Firefox.

[ssj12]

I disagree. There are many applications that are 64-bit. I have a few right now. Xfire chat, FireFox Nightly, Java, EVGA Precision, CoreTemp, Crysis, audio and video drivers, WinRAR, etc. There is no end to the number of already available applications and drivers available in 64-bit. Mozilla in general is being quite lazy since Safari has been 64bit for a while, and IE9 does have a 64bit option. Even Google has a 64bit version in the works.

The userbase is already there for 64bit, Mozilla just needs to capitalize on it, which they havent.

http://windowsteamblog.com/windows/b/bl ... ows-7.aspx

article is a bit dated, but based on a year later I could see percentage of 64-bit systems being much higher as most laptops and desktops sold are 64-bit now.

I'm sorry - the only applications there that actually could be useful with more then 3gb per process, or extra 64bit optimisations is WinRAR and Crysis - I highly doubt that Coretemp needs to be be 64bit apart from because you are using a 64bit OS - same with the drivers you mentioned.

I personally find FF unresponsive with anything over 1GB of RAM, so what good would be letting it have more then 4GB?

64bit Windows is way more stable then 32bit. Also I do a lot of photoshop work, 64bit speeds things up quite a bit compared to the 32bit version. With proper optimizations anything will run faster because they can allocate more resources to the ram versus hard drive cache or reloading from an HDD or even an SSD. While no, I dear god hope FF doesn't use over 3GBs of ram, but if I remember right 32bit windows limits the amount of ram use per process versus 64bit. I think the limit is like 1GB per process so that is the reason why your system/browser is chugging. Even with 1.5Gbs of ram being allocated versus 1 greatly help. And since a 32bit Windows can only read 3.7Gbs (including GPU ram) having a 64bit OS and 64bit software is a massive help overall on performance and usage of your system's capabilities.

Still more support is needed. If more people switched to using a 64bit OS, more developers would develop for it. They would be forced to switch to the architecture and be better prepared in ten or so years for 128-bit OSs.

[James]

So you want Mozilla to release a product, that isn't ready for prime-time, in order to compete with another product that is equally unready for actual use? That logic doesn't make any sense.

My logic makes perfect sense. For some reason, you are not following. I don't know where did you read that I wrote that I "want Mozilla to release a product, that isn't ready for prime-time". You are putting words in my mouth, and then say my logic doesn't make sense? Well, let's go on...

Not putting words in your mouth but informing you that it is indeed not ready for stable prime-time use yet as it is one of the reasons why there are no x86_64 release builds of Firefox for Windows yet.

[ssj12]

So you want Mozilla to release a product, that isn't ready for prime-time, in order to compete with another product that is equally unready for actual use? That logic doesn't make any sense.

My logic makes perfect sense. For some reason, you are not following. I don't know where did you read that I wrote that I "want Mozilla to release a product, that isn't ready for prime-time". You are putting words in my mouth, and then say my logic doesn't make sense? Well, let's go on...

Not putting words in your mouth but informing you that it is indeed not ready for stable prime-time use yet as it is one of the reasons why there are no x86_64 release builds of Firefox for Windows yet.

I think the main reason for no prime-time ready Win64 FireFox is lack of focus. But I have seen some promising pushes. The 64-bit tracking pages have been cut down from only about half complete to well over half complete over the last four or so months. I could see FireFox 7 or 8 being 64-bit.

[Bluefang]

I totally disagree with that. The browser should work with plugins that are built for the architecture that it was built for - 64 bit, in this case. If that is a "sine qua non" premise, then we are doomed - because we'll not have a Firefox x64 in a foreseeable time. Maybe after Google Chrome releases a x64 version (which probably won't work with 32 bit plugins, either). :wink:

One prominent reason for using a multi-process design (out-of-process plugins) is the ability to load binaries of differing architectures.

One of the goals while doing something like the 32bit-64bit transition is to make it as seamless to users ass possible. If you don't think that's the case, then kindly explain why every 64bit consumer CPU is x86_64, and why every consumer 64but OS runs both 32bit and 64bit binaries.

This whole 'seamless transition' is part of the reason why 64bit adoption has been so slow.

No kidding... And what version of Flash Player has no known security flaws? Obscenely out of date? Maybe, if Firefox x64 already existed, it wouldn't be so.
...
If " known security flaws in Flash Player" is reason to "highly discourage" its use, then all Flash Player versions use should be "highly discouraged"...

There is a severe difference between using the latest version of a piece of software, knowing that there are probably yet to be discovered exploits, and using an old/unsupported version that has well known exploits in the wild. Especially in such a high-risk environment like the internet.

No. Absolutely not. Plugin developers are taking their time and avoiding development costs because browser developers are reluctant (some irracional fear and lame excuses maybe?) to support 64 bit architectures - even when they already have done most of the work, like Mozilla - and when there are a large number of people asking for it. That's the real situation.

There is a drastic difference between a stand-alone application, and an application that, in part, depends on 3rd party components.

In order to present the 'full' internet, you need to use plugins. If a 64bit browser does not support loading all available plugins, then you are essentially neutering the internet, which makes using such a browser much less desirable.

This is very much a chicken and an egg problem. Should plugin providers develop plugins for a browser that doesn't exist? Or should a browser move to the new architecture and enjoy a compromised experience?

The rational mind would come to the conclusion that allowing a browser to load multiple plugin architectures is the best option moving forward.

My logic makes perfect sense. For some reason, you are not following. I don't know where did you read that I wrote that I "want Mozilla to release a product, that isn't ready for prime-time". You are putting words in my mouth, and then say my logic doesn't make sense? Well, let's go on...
...
Well, let's try it again. I wrote:

Releasing the Win64 Firefox would bring another (not small) advantage - the competition is IE x64, which is - well, weak. Or are we going to wait the release of Chrome x64 to start worrying about it?

What did I meant? That releasing Firefox x64 - meaning as soon as possible, after throwing the lame excuses out ( I NEVER wrote release it as it is) - they would drive a hard blow on their competitors.

So, let me get this right. You want Mozilla to release a browser with dodgy plugin support and known performance issues... in order to compete with a browser that has dodgy plugin support and known performance issues. Where does this make sense?

Releasing a crap product in order to compete with a crap product just leaves you with 2 crap products.

Since there's still no real advantage to using a 64bit browser (aside form memory usage), it seems like a better plan to do it right the first time.

And I'll add this: Mozilla needs to think seriously on getting back to the road of innovation.......

And the argument flies off into left field.



For the record, plugin support is not the only thing holding up an official release. IIRC there are performance regressions, compiler problems, lacking build/test infrastructure (i.e. Mozilla needs more hardware), and so on.

[mat--]

For the record, plugin support is not the only thing holding up an official release. IIRC there are performance regressions, compiler problems, lacking build/test infrastructure (i.e. Mozilla needs more hardware), and so on.

One of those issues is that currently Mozilla's buildbot produces Windows builds using MSVC2005. In order to properly support 64 bit builds they need to switch to MSVC2010. The switch is going to happen soon (http://blog.kylehuey.com/post/701537888 ... -msvc-2010), so we _might_ see Firefox 8 or 9 with "official" 64 bit support.

[Fabfire]

And I'll add this: Mozilla needs to think seriously on getting back to the road of innovation.......

And the argument flies off into left field.

Well, we'll have to agree that we disagree in most basic points of this issue.

[mikedl]

Wow, great discussion. Still, Bluefang, I side with Fabfire.

The sooner Mozilla releases an official Win x64 build, the sooner the demand for official Win x64 plug-ins will surface.

MS made the leap, initially - I believe that is the only reason Flash "Square" even exists at this moment for Win x64.

Seriously, 32 bit architecture is dying a fast death. It would be nice to see Mozilla at the forefront of welcoming 64 bit architecture on Windows.

I have very few users who even have Win7 32 bit installed these days.

Besides, Bluefang, you're on Linux x64, why would you care about Win x64?

[malliz]

My guess is the tipping point will come when there are substantially more 64bit machines in use than 32 bit, but at the moment unless you can prove that point has been reached then all the hyperbole is moot

[mikedl]

Well, XP is about at the end of it's lifespan - in my unofficial estimation, the remainder of the bulk of the Win32 user base.

All of my users are on 64 bit Win7 these days.

However, empirical evidence? I dunno. I only know with what I deal on a day to day basis that requires machines to run 64 bit AutoCAD and PhotoShop CS5.

Again, it would be nice to see Mozilla at the forefront of 64bit Win browser implementation.

Just my .02.

[watdis]

This goes back to the first point. Because plugin venders are reluctant to support 64bit architectures, browsers need to support 32bit plugins.

I disagree. There are many applications that are 64-bit. I have a few right now. Xfire chat, FireFox Nightly, Java, EVGA Precision, CoreTemp, Crysis, audio and video drivers, WinRAR, etc. There is no end to the number of already available applications and drivers available in 64-bit. Mozilla in general is being quite lazy since Safari has been 64bit for a while, and IE9 does have a 64bit option. Even Google has a 64bit version in the works.

I think you completely missed the point of what I said. I never said anything about Mozilla producing a 64bit browser. I just said that it needs to support 32bit plugins. And Mozilla agrees, which is why there hasn't been an official Win64 release. Arguing about priorities is a completely different matter.

FF is perfectly fine when it's using 2GB of RAM, 300 tabs. With Azure and electrolysis coming, I'm not sure if FF will be unresponsive ever, maybe only in extreme cases. With HTML5 and lots of media on internet in right now and in the future, I don't think powerusers will be fine with 32bit build of FF.

That's not a particularily convincing argument.

* Azure is just a rendering API that fronts OpenGL, DirectX, D2D, Cairo, or any other rendering backed they desire. Firefox already uses all of these rendering backends on the various platforms. Azure is just going to change how Firefox uses them.

* Electrolysis probably won't see that drastic of a resource overhead. It's not going to be process-per-tab (à la Chrome).

* HTML5. What? So sites are going to start using <audio> and <video> instead of Flash. That's a good thing, and will probably improve resource usage.

And, in this context, "power user" is largely becoming a euphemism for those who must run the latest and greatest just because it is shiny and new. A true power user would know that 64bit is useful in certain specific applications and use-cases. A browser is not [currently] one of them.

BTW, Bluefang, your extension (Status4Evar) works wonderfully on 64 bit Windows Fx.

Why wouldn't it? It doesn't have any binary components. If it wasn't working correctly, it would be due to a flaw in the Win64 version of Firefox.

The funny thing about using 32bit plugins in 64bit is, why do we need 64bit plugins then? By releasing x64 FF now without support for 32bit plugins, plugin devs will probably try to make x64 versions of plugins. Sure, 32bit plugin support is needed later on in x64 for those plugins that will never get updated or for old plugins, aka backwards compatibility, but not as a main point.

I'm not arguing to release x64 build the same way as x86 is released now, but maybe offer it as an option somewhere on the site, some sort of silent release with notice in red letters that internet experience may differ because some plugins are not compatible. That way, casual Joe won't know or need it, but people who are interested would be able to use it, instead of just nightly users.

About performance, I think you missed a point. Right now if a site is coded badly or just very heavy, it could hang FF altogether, but with e10 I think that would just hang the content, but not FF itself. Azure helps with rendering speed, so there will be less stuttering while loading/using some heavy content.

I meant HTML5 not just as video and audio tags, but all the complex JS and CSS that are being used nowdays. Also, I don't think Flash is going anywhere soon enough, so it will be still used along HTML5 and other stuff.

Then there's the point about using lots of plugins and extensions. With x64 build, I bet all the addon-holics will be happy to use even more of those. Improving memory usage is a good thing but using RAM that is actually idle is even better. If x64 can up the limit on the average addon amount that you can safely use without crashes or RAM caps, I'd say that is a good thing. FF is all about tailoring the internet how you see fit.