Jump to content



Photo

Anyone using Waterfox?


  • Please log in to reply
28 replies to this topic

#16 cork1958

cork1958

    Neowinian

  • Tech Issues Solved: 2
  • Joined: 04-October 02

Posted 24 January 2014 - 02:48

FWIW and slightly off topic, but seen this topic on front page again.. I just installed Cyberfox and it seems pretty darn good. Definitely quicker than Firefox ever seemed to be.

 

I see that reply that mentions pcxFirefox. Going to have to give that one atry also, on my 32bit machines! :)

 

Thanks for mentioning it, Max!




#17 AMPSV

AMPSV

    Neowinian

  • Joined: 13-September 03
  • Location: Sol 3

Posted 24 January 2014 - 02:54

Cyberfox's home page

https://8pecxstudios.com/

 


 



#18 Shiranui

Shiranui

    Iconoclast

  • Tech Issues Solved: 3
  • Joined: 24-December 03

Posted 24 January 2014 - 03:10

IceWeasel FTMFW!!!

Spoiler



#19 The_Decryptor

The_Decryptor

    STEAL THE DECLARATION OF INDEPENDENCE

  • Tech Issues Solved: 4
  • Joined: 28-September 02
  • Location: Sol System
  • OS: iSymbian 9.2 SP24.8 Mars Bar

Posted 24 January 2014 - 07:49

Are you sure?
 
Waterfox features:

  • Compiled in Intel's C++ Compiler
  • Intel's Math Library
  • Streaming SIMD Extensions 3
  • Advanced Vector Extensions
  • Jemalloc
  • Profile-Guided Optimisation
  • /O3 Switch
  • 100% Extension Compatibility
  • 64-bit Plugin Support
  • Future Proof!

Most of those are things normal Firefox does (Like PGO, SSE, AVX, jemalloc, optimisation flags, etc.), and that's ignoring stuff like "Future Proof" (Whatever that means, WoW64 isn't going anywhere any time soon)

I also don't think using the Intel Math Library would do anything other than make the binary larger, it's a specialised library (Meant for scientific applications), and unless the code is changed to take advantage of it, it probably wouldn't do anything (Really the only thing it could be used for is matrix handling, but they already use optimised code for that) Same with using the Intel compiler.

The only functional difference is that you can use 64bit plugins, but since the majority of plugins are 32bit that's not really a benefit. And again, that's ignoring the actual issues the 64bit builds have (Currently Mozilla can't even do proper testing of the 64bit builds due to issues with them, so while they might work for a specific workload a end user might have, they're not stable enough for every day use for a large portion of people)

#20 Demz

Demz

    Neowinian

  • Joined: 23-January 10
  • Location: Australia
  • OS: Fedora Linux
  • Phone: Samsung Galaxy S4

Posted 24 January 2014 - 08:24

i think if you want a native 64bit Browser thats worth using on a OS, use it on MACOSX or Linux. Windows isnt worth running 64bit Browser but eventually the way OS's are going to go there wont be any 32bit Variant of the OS . 64bit only . even RHEL7 there aint any 32bit



#21 Pupik

Pupik

    Neowinian Senior

  • Joined: 09-December 05

Posted 24 January 2014 - 09:08

Waterfox is made by Mozilla ;)

What? No it's not.



#22 kronckew

kronckew

    Neowinian

  • Joined: 12-August 09
  • Location: Glos., UK
  • OS: win7 x64 sp1

Posted 24 January 2014 - 09:44

Waterfox is made by Mozilla ;)

no, it's made by mr. alex, a student in cyprus who has been rightfully attending classes and doing waterfox in his limited spare time. the delay of v26 was due to his exams taking precidence as well as working out a few bugs in intel's compiler that stopped the code from working.

 

Most of those are things normal Firefox does (Like PGO, SSE, AVX, jemalloc, optimisation flags, etc.), and that's ignoring stuff like "Future Proof" (Whatever that means, WoW64 isn't going anywhere any time soon)

I also don't think using the Intel Math Library would do anything other than make the binary larger, it's a specialised library (Meant for scientific applications), and unless the code is changed to take advantage of it, it probably wouldn't do anything (Really the only thing it could be used for is matrix handling, but they already use optimised code for that) Same with using the Intel compiler.

The only functional difference is that you can use 64bit plugins, but since the majority of plugins are 32bit that's not really a benefit. And again, that's ignoring the actual issues the 64bit builds have (Currently Mozilla can't even do proper testing of the 64bit builds due to issues with them, so while they might work for a specific workload a end user might have, they're not stable enough for every day use for a large portion of people)

 

32bit firefox doesn't use any 64 bit optimizations. the 64 trial builds available in mozilla nightlies is not optimized and uses straight 32bit code. the 64bit pgo builds at mozilla are very dodgy bleeding edge trials of bug fixes. most of the other third party builds modify mozilla code in some way. mr alex uses the same code with just the mozilla mandated name/icon changes for the 'about' window. the intel compiler and optimizations should make a faster 64 bit version than the other 64 bit versions, none of which are as yet faster than the 32 bit versions because the code is written for 32 bit. compilers do not change 32bit optoimized code to 64 bit optimized code before compiling it.

 

and 32bit addons (extensions), like adblock plus, work prefectly well in 64 bit versions. they are javascript and bit size independent.

PLUGINS like java (which you should not need to install, it's still insecure), and flash do require the 64 bit versions, which work perfectly well.

 

What? No it's not.

 

correct. mr. alex is a private citizen with no big corporation behind him, no funding other than his own purse, and no staff beyond himself. i salute his efforts and the years of time he has devoted to all of us sharing his knowledge and efforts..



#23 The_Decryptor

The_Decryptor

    STEAL THE DECLARATION OF INDEPENDENCE

  • Tech Issues Solved: 4
  • Joined: 28-September 02
  • Location: Sol System
  • OS: iSymbian 9.2 SP24.8 Mars Bar

Posted 24 January 2014 - 10:15

...
32bit firefox doesn't use any 64 bit optimizations. the 64 trial builds available in mozilla nightlies is not optimized and uses straight 32bit code. the 64bit pgo builds at mozilla are very dodgy bleeding edge trials of bug fixes. most of the other third party builds modify mozilla code in some way. mr alex uses the same code with just the mozilla mandated name/icon changes for the 'about' window. the intel compiler and optimizations should make a faster 64 bit version than the other 64 bit versions, none of which are as yet faster than the 32 bit versions because the code is written for 32 bit. compilers do not change 32bit optoimized code to 64 bit optimized code before compiling it.
...


None of that makes any sense, you do know how compilers work, right?

#24 Trap

Trap

    Trap

  • Joined: 20-January 14
  • Location: United States
  • OS: Windows 8

Posted 24 January 2014 - 23:45

Pale-Moon seems to work better.



#25 +Boo Berry

Boo Berry

    Neowinian Ghost

  • Tech Issues Solved: 6
  • Joined: 26-March 05
  • Location: United States

Posted 24 January 2014 - 23:55

I've used them all. Honestly, I just go back to 32-bit Firefox as I see no benefit and they're all actually slower than vanilla Firefox (that and Waterfox has known issues with locking up it seems). If I want to use a 64-bit build, I'll use Nightly.



#26 +snaphat (Myles Landwehr)

snaphat (Myles Landwehr)

    Electrical & Computer Engineer

  • Tech Issues Solved: 29
  • Joined: 23-August 05
  • OS: Win/Lin/Bsd/Osx
  • Phone: dumb phone

Posted 25 January 2014 - 00:58

None of that makes any sense, you do know how compilers work, right?

I was wondering what he meant by writing code to be optimize for 64-bit also...

 

Regarding there Intel Math Library (MKL), it contains many ABI compatible interfaces for alternative libraries (libm, fftw, lapack, blas, etc.) so it's a drop in replacement in many cases. Here, the cases I surmise that could potentially benefit are signal processing (visual and audio) and core math function related. Though that isn't to say that Firefox is actually is written to make use of the former -- I have no idea, it could be, it could be not. I know Chrome is for audio support.

 

Also, regarding ICC, it does, in general, do a better job of generating optimized code: http://www.behardwar...hitectures.html (note: CL is MSVC)



#27 The_Decryptor

The_Decryptor

    STEAL THE DECLARATION OF INDEPENDENCE

  • Tech Issues Solved: 4
  • Joined: 28-September 02
  • Location: Sol System
  • OS: iSymbian 9.2 SP24.8 Mars Bar

Posted 25 January 2014 - 01:38

Hmm, I didn't think of FFT, I know they're using hand rolled matrix code though, which wouldn't benefit from the math library (And it's used a hell of a lot more)

As for ICC, I looked up some comparisons to GCC and Clang, and they were pretty much a wash. Micro benchmarks are really a bad judge though, I know Mozilla switched from GCC to Clang for the OS X builds, even though it was slightly slower overall, because it let them use updated SDKs, so speed isn't the be all and end all when it comes to deciding what compiler to use (They're also looking at the possibility of using Clang on Windows because it can use the native SDKs, and doesn't have issues that MSVC or GCC have, etc.)

#28 +snaphat (Myles Landwehr)

snaphat (Myles Landwehr)

    Electrical & Computer Engineer

  • Tech Issues Solved: 29
  • Joined: 23-August 05
  • OS: Win/Lin/Bsd/Osx
  • Phone: dumb phone

Posted 25 January 2014 - 02:03

Hmm, I didn't think of FFT, I know they're using hand rolled matrix code though, which wouldn't benefit from the math library (And it's used a hell of a lot more)

As for ICC, I looked up some comparisons to GCC and Clang, and they were pretty much a wash. Micro benchmarks are really a bad judge though, I know Mozilla switched from GCC to Clang for the OS X builds, even though it was slightly slower overall, because it let them use updated SDKs, so speed isn't the be all and end all when it comes to deciding what compiler to use (They're also looking at the possibility of using Clang on Windows because it can use the native SDKs, and doesn't have issues that MSVC or GCC have, etc.)

I'm a bit biased, but I think they probably should switch to LLVM/Clang even if it is slower, I've worked with it and it really has a nice architecture in comparison to classical compilers.

 

I also don't necessarily think compiler optimization advantages would translate into better real-world noticeable performance benefits either in this case. It's really hard to see how "better performance" is going to translate for an interactive application even if you do have better scheduling because, at the end of the day, you generally aren't running long compute intensive code. What you are running just has to be good enough to not lag in real time, for example audio processing, you don't really need a faster FFT library if you are already streaming audio in real-time. Contrast that with HPC where you have applications that are running for days or months in some cases and the 20% is a large increase in performance there.

 

The problem with LLVM is that it historically has been fairly unsupported on Windows (that's started to change as of late last year) so I don't know if it is particularly viable at this point, it's definitely worth unifying the build environment of FF in the long run though for the project. I love that we are seeing pre-compiled builds now _finally_ though. 



#29 The_Decryptor

The_Decryptor

    STEAL THE DECLARATION OF INDEPENDENCE

  • Tech Issues Solved: 4
  • Joined: 28-September 02
  • Location: Sol System
  • OS: iSymbian 9.2 SP24.8 Mars Bar

Posted 25 January 2014 - 02:51

Yeah, the fact that Clang can use the native Windows SDK without changes is a huge benefit, GCC can't and MSVC has issues (Funnily enough, MSVC can't compile 64bit Firefox properly, that's what's stopping them from doing testing, they need to upgrade to VS2012)

And yeah, LLVM has quite a nice architecture, been a while since I played around with it much though.



Click here to login or here to register to remove this ad, it's free!