KDE & Xfce Don't Lead To Performance Wins Over Windows 8

[brosis]

Remember, the Linux driver is completely opensource.

Now compare it to fglrx vs radeon.

I say, Intel does a magnificent job, hats off - the driver I am and will be buying hardware for.

And much thanks to Michael!!

[curaga]

Haha, Michael pre-answered half these questions and still it seems many didn't RTFA

[Khudsa]

This is what I say: "?ZAS! ?En toda la boca!"

Thanks Michael!

[philip550c]

Thanks Michael

Seems like people have been giving you a hard time with every benchmark you have posted lately.
I think were getting cocky having steam for linux now.

[liam]

hah damn Michael put the whiner's in their place. I personally was interested to see these tests myself, though I didn't expect KDE would have performed better than Unity or XFCE. I was surprised to see how poorly it performed.

I think what would be a nifty test is to see how Windows 8 performs when using DX and GL versions of the same games (can be a different collection of games from the tests performed here though).

From his previous tests, I knew KDE wasn't going to do well, that's why I didn't mention it. However, xfwm has done well, and the even lighter WMs like openbox have been the fastest overall. Thus, the reason why I'd be curious about running these games sans any WM.

[ChrisXY]

but from other side, there is also almost no point in testing such powerfull games as serious sam 3 with intel graphics, which isnt made for performance; so we would see no difference, it would be like "max: 2 fps, min: 1 fps"

Actually, if you put it on low graphics details, it works quite smooth, but of course doesn't look very good.
Also, the validity of one game benchmark would be questionable anyway as it may or may not be more or less well optimized and graphics drivers may or may not make something of it. Take the beta update from today as example.

- Thoroughly refactored and optimized postprocessing shader system. Some effects are up to 5x faster.

But back to topic. I think I was the first to bring that old topic up again.

But first about this:

On a quick glance this title looks like it is saying windows 8 has better performance than linux. Could you use titles like e.g. "Intel driver for windows 8 outperforming intel driver for linux" instead, that better describe what the article wants to say?

You actually made it worse. This is from this new article:

When publishing the OpenGL performance results yesterday showing Windows 8 generally leading with a performance advantage over Ubuntu Linux, there was the usual large portion of the Linux community in disbelief.

No, the intel driver for windows was leading with a performance advantage over the intel driver on ubuntu linux.

I mean, what's up with that title?

KDE & Xfce Don't Lead To Performance Wins Over Windows 8

That sounds like a really broad statement but it actually is only about the intel driver.


And regarding the compositing I was just asking for confirmation that compiz/unity doesn't make the performance worse:

So, this was done with unity's/compiz' compositing. There shouldn't be a noticeable difference for the open source drivers between running opengl on a composited desktop vs on a non-composited one anymore, right?

I realize that some other people jumped to the conclusion that compiz/unity must make the performance worse. But I just wanted to see it. Pretty much this benchmarks combined with the previous article was all I wanted to see.

I mean, people don't just randomly request that you compare it against non composited desktops, your benchmarks have shown it to be a significant influence in some benchmarks in the past. Glad that time seems to finally come to an end though.

[rikkinho]

intel gpu

i have better fps on steam counter strike condition zero and 1.6 with intel hd3000 than i have with fglrx 13.2 amd hd 6770m (hybrid graphics laptop), intel opensource driver are in good shape but ofc needs more a 1\2 years to match windows perfomance

[Kano]

Basically those benchmarks are a bit boring - you test even Nexuiz then everybody knows that this game nobody plays anymore. Would be much cooler when you test current games from Steam. For 3 games i have got i wrote scripts, basically one is very generic and should work with any Source engine game. All you need is to record your own demo(s). For Killing Floor you can bind keys for that and for TF2 and other Source engine games you enable the developer console in adv. keyboard settings and use "record demoname" and later "stop". The Source engine has got some nice autocomplete features and you can use timedemo/demoinfo directly in the console. The result(s) will be saved in a SourceBench.csv which also shows the gfx driver used and other settings - so in theory you dont need a script for that. I wrote some example code however to help integration in PTS later, but i need at least an option that can reset the config to a specified default for the Source engine, maybe i replace the current config file when there is no override option. A valve dev could help there... So here are 3 example scripts that let you select a demo and then runs the game (for Killing Floor it does not even require to have Steam running, would even work with a copy that is left on hd from a guess pass) as long as you have got Demo(s) in "~/.killingfloor/Demos".

Killing Floor:

http://kanotix.com/files/fix/tmp/kf/bench-own-demos-killing-floor.bash

Then for Source engine games that use the steampipe system which are always in the common subdir like Half Life 2 Deathmatch:

http://kanotix.com/files/fix/tmp/hl2mp/bench-own-demos-hl2mp.bash

And for old style Source engine which load EXTREMELY slow like Team Fortress 2 (without Beta):

http://kanotix.com/files/fix/tmp/tf2/bench-own-demos-tf2.bash

Basically the last one can be use for every Source engine game, all you need to adjust are 3 lines which should be really easy. But as you get a csv anyway i want at least Source engine benchmarks...

[pandev92]

Serious sam BFE3 under linux with intel hd4000 is unplayable, 3-4 fps, but under windows 8 with the same chipset, I can run this game on the mid detailes on 1920x1080 and I get 30/40 fps...

This is the very difference, intel drivers on linux, are sooo sloowwww on 3d games.

[rikkinho]

??

with 9.1.1 mesa? i don t have this problem =S with mesa 8/9 the steam are slow no dobt about that