WebXPRT 4 performance scores

[SchumannFrequency]

This seems to be a pretty poor CPU testing benchmark it fails to push all threads to anywhere near their maximum and very poorly multi core aware wit the max core's/threads being used is ~4 but on average it's mostly single core/thread leading to poor utilization of ~14% at most

According to the authors of the test, this is the most reliable browser benchmark.

It has been known for some time that browsers are single threaded. However, that doesn't mean it's a bad benchmark.
Even for Cinebench, many reviews often mention the single-thread performance in addition to the multi-threading performance.

Most apps that the average user uses every day are single-threaded.
Think of an email client, calculator, PDF viewers, file managers, text editors, browsers, LAME, Flutter (out-of-the-box), Telegram, etc.
For LibreOffice it is only Calc that uses multi-threading, for example LibreOffice Writer and Impress (which are much more commonly used than Calc) are both single-threaded.
One of the most distinctive features of Node. js is its single-threaded architecture.
Most Linux/Unix terminal tools are single threaded as Bash used to be single threaded.
Even the most popular backup tool of Linux systems (rsync) is single threaded.

It is very logical to conclude that single-threading benchmarks are what determine the average person's daily performance the most.

There are also certain advantages that single-threaded apps have over multi-threaded apps:
- less complex software build is time-saving for developers
- easier to debug = more stable software
- Intel Hyperthreading and AMD Simultaneous Multithreading (SMT) both have many vulnerabilities, meaning single-threaded apps score better in terms of security

PCManFM (Qt version) is probably also single-threaded and starts on OpenBSD in 0.1x second.
It's not like multi-threading is going to make this app noticeably faster for standard users. In many situations multi-threading is of no use.

lximage-qt and many other image viewer apps don't seem like multi-threaded apps to me either.
But lximage-qt opens in exactly 0.1 second on OpenBSD with a slow SATA SSD.

It appears that Firefox will use multi-threading for some tasks in the future.

Servo Web Engine Now Leverages Multiple CPU Cores For Rendering HTML Tables - Phoronix

www.phoronix.com

Suppose all apps were written in the programming language that is most optimal for multi-threading (Haskell), the Intel 12700KF could be faster than the 7800X3D in almost all apps.

Because it's Haskell, developers also save a lot of time.
The app is also more secure than in other programming languages since Haskell apps have significantly fewer bugs than Python/C/Java/C++/JS apps.

 

[SchumannFrequency]

Hardware: Intel 12700KF (stock) -- G.SKILL RIPJAWS @3600 CL18 (stock) -- Sapphire RX 7600 -- ASRock B760M-ITX/D4 WiFi -- fractal design DEFINE NANO S -- bequiet! SYSTEM POWER 10 550W -- DeepCool AG500BK ARGB -- EVO 850 500GB
Software: OpenBSD -current, bspwm, Mesa open-source driver, UFS file system, Chromium 127.0.6533.72

In post 20 on this thread, I arrived at the following result:
Chromium + Gentoo (KDE Plasma) + Intel 12600KF + DDR5 @ 6000MHz = 428 (WebXPRT 3)

Chromium is normally going to feel very similar on Gentoo as it does on OpenBSD -current in terms of performance.

 

[BlaezaLite]

 

[SchumannFrequency]

Hardware: Intel 12700KF (stock) -- G.SKILL RIPJAWS @3600 CL18 (stock) -- Sapphire RX 7600 -- ASRock B760M-ITX/D4 WiFi -- fractal design DEFINE NANO S -- bequiet! SYSTEM POWER 10 550W -- DeepCool AG500BK ARGB -- EVO 850 500GB
Software: Clear Linux, Gnome, Mesa open-source driver, EXT4 file system, Firefox 128.0.2

This result is without using the new schedutil patches. It could be that these new patches lead to higher performance in this test.
Source https://www.phoronix.com/news/Schedutil-30p-Speedometer-Boost

 

[AVATARAT]

9700x@PBO
I don't know what's the Mozilla problem with this bench.

Waterfox

 

[SchumannFrequency]

I don't know what's the Mozilla problem with this bench.

I don't consider Waterfox to be Mozilla but a mod of a Mozilla app.
Firefox on Linux performs similarly and sometimes better than Chromium-based browsers in WebXPRT 3/4.

Can you test Firefox Nightly?
It's the bottom option on this page: https://www.mozilla.org/en-US/firefox/channel/desktop/

Also, can you do a multi-core benchmark of the Ryzen 7 9700x?
For example, the test below gives a good idea of the multi-core performance.

Blender 4.0.1 BMW27 demo test

Render time: 119.2 seconds This test is famous in part because Intel or AMD have frequently used the test to demonstrate the advances of their latest-generation CPUs. Hardware: Intel 12600KF (stock) -- Kingston 6200 MHz CL36 -- Sapphire RX 7600 -- BIOSTAR B760MZ-E PRO -- Antec P6 -- Xilence...

www.techpowerup.com

 

[AVATARAT]

Can you test Firefox Nightly?

Here's the difference day and night
My Waterfox has too many extensions and limitations, so they probably make the result worse.

 

[lolek86]

New 6000MTCL26 profile

 

[sophie]

14900ks

 

[Paramount]

 

[BNSoul]

9800X3D mostly stock

 

[phanbuey]

OC 9800x3d with a VM running in the background

 

[SchumannFrequency]

https://i.ibb.co/WpKS99g3/2025-05-01-152134-1920x1080-scrot.png