Jump to content

Force "multiprocess mode" in FF 52


Recommended Posts


5 hours ago, VistaLover said:

A fourth (if ipc.processCount;1) process can be spawned if the pref browser.tabs.remote.separateFileUriProcess;false is toggled to true... 

What is true for Fx 53.0 should also apply to St55 under Win7!

Interesting! At first, toggling that pref to true had no effect on my Win7 system; with processCount set to 2 and 2 tabs open, I had 4 processes before, and after, toggling separateFileUriProcess and restarting.

But then I read the pref name closely,  opened an HTML file on my own PC, et voila! A fifth process appeared. Closing the tab showing that HTML file dropped the process count back to 4.

I don't know that it's necessarily worth toggling that pref, even on Win7, but it's another surprising feature of this browser.

6 hours ago, VistaLover said:

Simply changing the value of the app.update.channel pref doesn't change already installed code

Yeah, I know; but what I didn't mention originally is that I actually looked at the XP-blocking code at the Bugzilla link you provided earlier:

#if defined(XP_WIN)
  if (Preferences::GetDefaultCString("app.update.channel").EqualsLiteral("release") &&
      !IsVistaOrLater()) {
    gMultiprocessBlockPolicy = kE10sDisabledForOperatingSystem;
    return gMultiprocessBlockPolicy;
  }
#endif

(This was back around FF 48.) As you can see, it uses an #if defined(...) block to check whether the build targets Win XP, but it does a run-time check of the app.update.channel pref rather than using another #if defined(...) block. Obviously the code has been changed since then - otherwise e10s would work on Win XP in FF 52 ESR since it's not a "release" build - but I figured they probably still did a run-time check, so changing the pref was worth a try. However, it didn't work.

Of course I don't know if building the browser as "default" vs. "release" or "esr" would let e10s be enabled the "normal" way. I'll leave that as an exercise for others. ;) As long as there's another way to enable e10s without recompiling the browser, I'm happy; just very curious.

Link to comment
Share on other sites

I tried the acceleration test which @mockingbird linked to, and my Firefox on XP only makes 20fps!
I guess no hardware acceleration then!
:(
As an off-topic aside, I tried the test in Windows 10 as well, and was surprised that 64 bit FF 66 only made 60fps.
IE11 and Edge both made well over 200fps on the same test!
:o

Link to comment
Share on other sites

3 hours ago, Sampei.Nihira said:

Have you checked whether the changes cause problems with the plugin-container.exe?

With XP....

Well, Silverlight and Java tests seem to run OK; I realize that's not a very thorough test, but all plugins use plugin-container.exe AFAIK.

1 hour ago, Dave-H said:

I tried the test in Windows 10 as well, and was surprised that 64 bit FF 66 only made 60fps.
IE11 and Edge both made well over 200fps on the same test!

That's weird. I got 181 fps on 32-bit St 55 (which AIUI is basically FF 53) on 64-bit Win 7. I wonder if the older FF might outperform the newer one on Win 10?

Next week I'll try it on both the Win 7 and XP "sides" of the same PC and see how much difference HW acceleration makes.

Link to comment
Share on other sites

Usually it is with Flash that you can have some problems.

Flash is always run out-of-process.

With Process Explorer version 16.12 it is possible to highlight what has been said.

ZhvaOdIW_o.jpg

It is also interesting to note:

- The "Description" reads "Plugin Container for New Moon" (not Pale Moon) and the "Company Name" is "Mozilla Corporation".

- Both PaleMoon.exe and Plugin-container.exe are not signed.

- Plugin-container.exe also injected (by default) the MBAE dll.

Edited by Sampei.Nihira
Link to comment
Share on other sites

On 5/11/2019 at 11:26 AM, Dave-H said:

I tried the acceleration test which @mockingbird linked to, and my Firefox on XP only makes 20fps!
I guess no hardware acceleration then!
:(
As an off-topic aside, I tried the test in Windows 10 as well, and was surprised that 64 bit FF 66 only made 60fps.
IE11 and Edge both made well over 200fps on the same test!
:o

This did work at one point on XP and was disabled...  I don't remember the last version when it worked, but it was a long, long time ago...  Maybe like a twenty-something version.

This is something that I genuinely believe can be hacked back into Firefox on XP.  SvyatPro did some nice work hacking it into Chrome on XP, and it worked well.  I can send you the details if you're interested.

Link to comment
Share on other sites

17 hours ago, Sampei.Nihira said:

Usually it is with Flash that you can have some problems.

Flash is always run out-of-process.

I have Flash but leave it disabled unless I come across a site that needs it, so I haven't really tested it with e10s.

17 hours ago, Sampei.Nihira said:

Both PaleMoon.exe and Plugin-container.exe are not signed.

Not surprising in the least. Code-signing certificates cost money and @roytam1 is building these browsers for free. I'm sure he'd be happy to sign them if you ponied up to buy him a cert. Or you could go with MyPal if their .exe files are signed. AIUI there's very little difference between MyPal and NM (mostly just which of MCP's builds are used).

BTW I don't think e10s even works with NM; only with Serpent.

Link to comment
Share on other sites

On 5/11/2019 at 10:26 AM, Dave-H said:

I tried the acceleration test which @mockingbird linked to, and my Firefox on XP only makes 20fps!
I guess no hardware acceleration then!
:(
As an off-topic aside, I tried the test in Windows 10 as well, and was surprised that 64 bit FF 66 only made 60fps.
IE11 and Edge both made well over 200fps on the same test!
:o

 

On 5/11/2019 at 12:00 PM, Mathwiz said:

That's weird. I got 181 fps on 32-bit St 55 (which AIUI is basically FF 53) on 64-bit Win 7. I wonder if the older FF might outperform the newer one on Win 10?

Next week I'll try it on both the Win 7 and XP "sides" of the same PC and see how much difference HW acceleration makes.

Big difference: 15 f/s on XP vs. 93 f/s on 7, using Serpent on the same machine. (I'm a bit surprised that my home machine is almost twice as fast as my work machine at this particular task, but whatever.) I assume most of that is due to not using hardware acceleration.

Link to comment
Share on other sites

Well I tried Firefox 66 on Windows 10 again, and still get only 60fps on the test.
It's set to its default automatically determined configuration, with hardware acceleration on and a maximum of 8 processes (which is the default).
It's still massively out-performed by IE11 and Edge.
Off topic for this thread of course, but a very strange anomaly.
:yes:

EDIT: Actually I've now done some more tests, and sometimes FF 66 on Windows 10 goes as high as 170fps, but I'll then try again and it's around 40fps!
No consistency to it at all.
Anyway, off topic so I'll concentrate just on FF 52 on XP from now on, I promise!
:lol:

Edited by Dave-H
Correction & Addition
Link to comment
Share on other sites

2 hours ago, Dave-H said:

Anyway, off topic so I'll concentrate just on FF 52 on XP from now on, I promise!

:lol:

If you feel like testing later operating systems and browsers, test Chalkboard.

IE11 is by far superior to anything else in terms of raw speed...  Microsoft probably has all sorts of undocumented API calls and under-the-hood secrets that they keep from devs like Chromium, or the Chromium devs just don't care enough about Windows to match its speed.  IE11 also beats Edge, hands down.

On my old Windows 10 Tablet which I use in the bathroom, I exclusively use IE11...  It's the only browser I can use in a reasonable manner on such a slow system.

Link to comment
Share on other sites

Interesting test.
Google Chrome 49 managed it in 11.16 seconds.
Opera 36 managed it in 11.76 seconds.
Firefox 52 ESR managed it (eventually) in......... wait for it......... 380.18 seconds!!
I thought it was never going to finish on Firefox. It took over six minutes!
Something must be seriously amiss here surely?!
:o
 

Link to comment
Share on other sites

2 hours ago, Dave-H said:

Interesting test.
 

Yes, it is.

Without forcing "multiprocess mode" in Firefox 52 on XP:  anywhere from 16 seconds to 38 seconds.

Google Chrome Portable on XP:  5.51 seconds.

IE 11 on Windows 7:  5.27 seconds.

Link to comment
Share on other sites

I know I promised i wouldn't mention it again, but just quickly for comparison, Firefox 66 on Windows 10 managed it in 44.79 seconds.
Much much better than FF 52 on XP, but still pretty bad compared with it seems all other browsers!
So, it looks like a general problem with Firefox, all versions seem to underperform badly on these tests compared with other browsers.
:dubbio:

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...