Jump to content

Why are browsers so slow when running under XP?


Snear

Recommended Posts

I am using roytam1's great browser builds and I am always wondering, why are they running so slow when using XP. The same browser build used with Win 7 (32 or 64 Bit) runs much much faster. Ads are not the problem, neither is the agent string used. Even if there is nothing I can do about it, I am interested in the reason. Here is one site which is free of ads and shows up pretty quickly in Windows 7 with roytam1's KM-Goanna, Basilisk and PalemoonXP, but won't even load completely in Windows XP: https://www.binance.com/de/trade/BTC_USDT

PS: No discussion about crypto's please.

 

Edited by Snear
Link to comment
Share on other sites


I suffer from this slowness with Yahoo mail since January 2019.... 

I have even tested it under 360 EE or Maxthon but the problem persists.

However google Gmail is pretty fast and smooth with any browser. :dubbio:

 

 

Link to comment
Share on other sites

Mozilla browsers are terribly unoptimized out of the box. Hence why they run so horribly. Use the UOC Patch and Enforcer, it will make a difference.

Depending by your hardware setup, you might get the fastest speed with a patched Serpent/Basilisk 52 according to the tests I have done in the latest version which I still have to release.

On a very old Asus L3000D laptop with an Athlon XP 2500+ and a SiS Real256 IGP (DirectX 7 only graphics chip), patched Roytam1's Serpent 52 scrolls incredibly fast and smooth, on par with Internet Explorer, making it my browser of choice on that laptop. Which patched browser will run faster, depends by your hardware configuration. As always, YMMV. Meanwhile you can use the currently public build, N2M.

Edited by looking4awayout
Link to comment
Share on other sites

Get the same problem with the latest New Moon (palemoon-28.9.0a1.win32-git-20200125-eb49b28df-uxp-a64ac6d70-xpmod.7z) and previous versions when run on XP on fast Core 2 Quad.

Is the UOC patch needed with Roytam latest New Moon builds (when using C2D or newer processors) or is it already optimized?

Link to comment
Share on other sites

Depends by the stepping of the C2D and the graphics card/IGP. I have used it successfully on an Acer laptop with a Merom C2D and an Intel GMA that was too slow to run the foxboxes at stock settings.

I'm lately testing the new version of my patch, still unreleased at the moment, on an Asus EeePC 4G (Celeron M 900Mhz overclocked to 1,1GHz, 2GB of RAM and an Intel GMA 900, overclocked with GMAbooster), and it runs very well, despite the many shortcomings of the CPU. I'm using NM27 and K-Meleon Goanna as browsers there. 

The UOC Patch is not included with Roytam1's browsers, you must download it separately.

Edited by looking4awayout
Link to comment
Share on other sites

On 1/27/2020 at 10:38 PM, Snear said:

The same browser build used with Win 7 (32 or 64 Bit) runs much much faster.

''Performance-wise, XP is also not that good anymore. Windows Vista and especially 7 make much better use of multicore CPUs, and large amounts of memory.''

https://scalibq.wordpress.com/2012/04/17/windows-xp-the-gold-standard-of-windows-oses/

Link to comment
Share on other sites

On 1/28/2020 at 9:14 AM, looking4awayout said:

Mozilla browsers are terribly unoptimized out of the box. Hence why they run so horribly. Use the UOC Patch and Enforcer, it will make a difference.

Even if they are - it is still unclear to me, why the same browser (same executable!) runs so much faster on WIn7 x86 than on Win XP Pro. It has to do something with Windows XP. I never recognized this behaviour for other software. It might have something to do with memory management, network traffic or network connection. I was hoping that someone knows, but it seems it is kind of a mystery.

 

Link to comment
Share on other sites

On 2/1/2020 at 8:23 PM, Snear said:

Even if they are - it is still unclear to me, why the same browser (same executable!) runs so much faster on WIn7 x86 than on Win XP Pro. It has to do something with Windows XP. I never recognized this behaviour for other software. It might have something to do with memory management, network traffic or network connection. I was hoping that someone knows, but it seems it is kind of a mystery.

 

Something strange with these browsers in XP (firefox, new moon etc). After so many tabs being open (not a lot) I get full 99-100% CPU usage in task manager and intermitten freezes and puases. The computer hasn't crashed just the browser using 100% cpu usage.

Trying to analyze and find the cause using Sysinternals 'Process Explorer' doesn't help. This is the same when using a clean XP install with just SP3 or a clean install with all posready 2009 updates.

Exactly the same browsers with win7 and no problems.

Link to comment
Share on other sites

I haven't noticed much difference between Windows 2000 and Vista/7/2012R2/10. XP should be equal if not better (unless BWC's kernel is more optimized than XP's), and 2003 definitely has some scheduling improvements over 2000 which allow me to listen to a YouTube video while having MSFN open (in win2k the audio becomes garbled on a Core Duo T2400).

Vista with Platform Update and above also have DX11/D2D/DirectWrite for hardware acceleration IIRC, which NT5 lacks, except for Windows 2000 with the BWC-provided wineD3D DX10/11 module possibly. In fact the site in the OP's link seem to make use of the DX10/11-type acceleration as they crash the browsers on my workstation as a result of OpenGL calls being directed to nss3.dll (instead of opengl32.dll). So that may be the missing ingredient for NT5.

Edited by win32
Link to comment
Share on other sites

On 2/5/2020 at 7:13 PM, ricardrosen said:

Something strange with these browsers in XP (firefox, new moon etc).

Firefox browsers have never been optimized enough for XP. :crazy:

with Chrome I can go beyond RAM + pagefile (virtual memory) limit with no issues @daniel_k

while on Firefox if you open-close too many tabs it might crash at any time.

just in china there are almost 300 web browsers and most of them are based on chrome @roytam1

although roughly 4 browsers do meet our requirements = up to date, compatible with XP 32, full UI in english,

two out of them are well known here Maxthon and 360chrome

''chupacabra'' may want to give the two other browsers a try (based on chrome 75) @ED_Sln

 

 

Link to comment
Share on other sites

It shouldn't be that bad as the OP describes due to an OS. Things render a bit faster on newer OS for me, one apparent example being HexGL game for instance which runs choppy on XP (x64) for me, even though ANGLE is disabled and NVIDIA's drivers are generally good at OpenGL. There's a difference in how fast a web pages appear, albeit not dramatic.

The biggest performance killer on XP is running the browser on secondary screen. Rendering is normally off-loaded to GPU and DirectX on XP has an awful characteristic that when application's window in which it's used is moved to another screen makes some slow copying process going on constantly in the background, slowing everything down.

Using Mozilla browsers, it's possible to hack xul.dll to call IDirect3D9::CreateDevice with a different adapter number, binding the browser to another non-primary display, so it will run fine on the one that was chosen (but not the others). With later OS (Vista+), it doesn't matter which display it's bound to, the parameter is only relevant for fullscreen mode.

Maybe it would be possible to make it work right on XP with multi-mon setup using DirectX, but it might not be trivial task. There's a pref to use OpenGL (which doesn't have that problem) instead of DirectX, but it seems broken on Windows (crash). At very least, @roytam1 could add a workaround in form a pref setting to select adapter/display number passed to IDirect3D9::CreateDevice.

On 2/6/2020 at 6:27 PM, win32 said:

Vista with Platform Update and above also have DX11/D2D/DirectWrite for hardware acceleration IIRC, which NT5 lacks, except for Windows 2000 with the BWC-provided wineD3D DX10/11 module possibly. In fact the site in the OP's link seem to make use of the DX10/11-type acceleration as they crash the browsers on my workstation as a result of OpenGL calls being directed to nss3.dll (instead of opengl32.dll). So that may be the missing ingredient for NT5

How did you come to the conclusion that OpenGL calls are made? Also, not sure how rendering calls could be made to the library dealing with cryptography and such things. That site doesn't use WebGL neither.

Link to comment
Share on other sites

On 2/8/2020 at 2:25 PM, UCyborg said:

How did you come to the conclusion that OpenGL calls are made? Also, not sure how rendering calls could be made to the library dealing with cryptography and such things. That site doesn't use WebGL neither.

I had it the wrong way around, as nss3.dll was making calls to various libraries, including libegl.dll, kernel32.dll and palemoon.exe, but usually not the right one, opengl32.dll. "glGetPointerv" was one such call. These calls were only made when loading that site on a system with the DX10 components; no OpenGL calls are ever made when the module is not installed, nor does it make them on most other sites even when installed.

I used Dependency Walker to evaluate.

UPDATE: Tested on Vista SP2 + Platform Update and mostly the same behaviour as on Windows 2000 with DX10 module, but it loads successfully.

I also compared loading times on XP x64 and 2012R2 (Serpent 2020-02-08 and Pale Moon 28.8 respectively, all x64). They were the same on my Xeon X5670 machine at 13 seconds. So hardware acceleration differences are indeed not a factor. But the site is quite sluggish in general.

Oddly enough, if you tell the browsers that they're on Vista or 7, the pages load without crashing in win2k.

Edited by win32
Link to comment
Share on other sites

I use fcwin2k to "tell" New Moon 28/Serpent 52 that they are running on a newer OS so that pages like the one listed in the OP load. This is with Windows 2000 with Extended Kernel v3.0e and with DX10 compatibility enabled in the extended kernel installer.

Edited by win32
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...