Browser benchmarks: Linux vs Windows

[Tom Mann]

nvidia-glx-new from the repository

This will be why - nvidia have been working to fix their drivers after a regression that hit 2d performance in KDE, maybe we're missing something on the render speed of Gnome too? Phoronix is our friend here.

Also, could you repeat the test with the latest alpha? We should look forward as well as back...

[zenkaon]

I agree strongly that something needs to be about X.org, it is far too slow.

At work I hook up my laptop to an external monitor, keyboard and mouse and then click a little button I have which runs

xrandr --auto

The performance of my laptop slows significantly when hooked up to the monitor, especially firefox.

I have a Dell Latitude D420 (2x1.2GHz 1GB RAM) Using repository intel drivers. Firefox 3.5.2. Debian Lenny (I know this is an ubuntu formum.....but ubuntu is tweaked Debain) X.org version 1.4.2

My Benchmarks for peacekeeper are:
Stand alone laptop = 562
Docked laptop = 246

OK so my performance isn't great anyway, but I am taking a 44% hit in performance by changing my 12" laptop screen to a 19" monitor.

[zenkaon]

Wow,

based on these peacekeeper benchmarks I suck google chrome on my laptop. I've only tested while docked so far, I'm sufficiently impressed and unless I see any hickups I've changed browser.

Firefox 3.5.2 just laptop = 562
Firefox 3.5.2 Docked to 19" = 246
Chrome 4.0.203.2 Docked to 19" = 1443

Goodbye firefox

[SunnyRabbiera]

Benchmarks should not be counted on as it really depends on your hardware.

[Nepherte]

Except this cannot be the case here, because we're using the same software. Unless you're saying all of them intentionally made those particular benchmarks score higher only on Windows. I find that unlikely.

You're forgetting the fact that browsers rely on remote procedures as well, inherent to the operating system. It is practically impossible to test the performance of the single application. Not to mention that the execution time can be separated in user time + system time.

You're also underestimating the value of scoring high in performance benchmarks. Companies would do everything to score high and hence get better sales. So yeah, a lot of companies intentionally forge better scores.

[chriskin]

sorry to say what is supposed to be said here
but using the sunspider (http://www2.webkit.org/perf/sunspide...sunspider.html) benchmark i found out that the same browsers are slower on my brother's Too-new-to-be-true ultra powerful desktop pc (wired connection)that runs on vista than my not-so-new laptop (wireless) that runs on ubuntu

so i just can't accept that chromium is faster over there than here

[unimatrix]

This will be why - nvidia have been working to fix their drivers after a regression that hit 2d performance in KDE, maybe we're missing something on the render speed of Gnome too? Phoronix is our friend here.

Also, could you repeat the test with the latest alpha? We should look forward as well as back...

I suspected something like this, yes.
I shall repeat the test tomorrow with the latest drivers.

[coldReactive]

I just got 1683 in Peacekeeper on 3.5.3 Firefox myself. Then again, I have a 3.0 GHz Duo Core and 6 GB of RAM and a GTS 250 Graphics card with 1 GB Video RAM.

[rookcifer]

A lot of it has to do with how the browser was compiled. Obviously, Windows and Linux use different compilers with different settings, etc.

I have seen benchmarks (on Phornoix I think) where it was shown that running Firefox in WINE was faster than running native Linux Firefox. This shows there is some difference in how the code is compiled.

[unimatrix]

A lot of it has to do with how the browser was compiled. Obviously, Windows and Linux use different compilers with different settings, etc.

I have seen benchmarks (on Phornoix I think) where it was shown that running Firefox in WINE was faster than running native Linux Firefox. This shows there is some difference in how the code is compiled.

I've also benchmarked Firefox native and Firefox via Wine get about the same results. The one on Wine is slightly lower. Don't really care what Phoronix says, the fact remains that Firefox on Linux is still very unresponsive and much slower.