Gentoo vs freebsd

blueworm · l33t Joined: 09 May 2003 Posts: 962

Was bored this weekend and decided to give freebsd a go.
I ran glxgears and to my supprise 1000 fps more than gentoo.

My system:
A7N8X K7 3000+
512Mb kingston HyperX 2225
GeforceFX 5900

Gentoo 2.6.2 vanilla kernel.
Nvidia 5336 driver. default depth 24 1200x1024
glxgears default window ~6200 fps

freebsd 5.1
glxgeras default window ~7200 fps

It gets worse

quake3 timedemo 1;demo four
Gentoo ~260.4 fps
freebsd ~301.8 fps

Can somebody please explain this to me...

Gentii · Guru Joined: 01 Feb 2004 Posts: 306

freebsd rules?

blueworm · l33t Joined: 09 May 2003 Posts: 962

jrz · Posted: Mon Feb 09, 2004 9:40 pm Post subject:

perhaps you should go through the source code and see what the differences are!
_________________
Press F1 for Help.

bleakcabal · Posted: Mon Feb 09, 2004 11:13 pm Post subject:

Maybe freebsd has less process, apps and stuff running all the time in the background ?

Duty · Apprentice Joined: 15 Nov 2003 Posts: 253

Good thing some of us are working on Gentoo/BSD, then

Trejkaz · Posted: Tue Feb 10, 2004 12:50 am Post subject:

I gather the same version of the video driver was being used on the BSD system... you didn't mention whether that was the case.

blueworm · l33t Joined: 09 May 2003 Posts: 962

Doikor · Posted: Tue Feb 10, 2004 7:32 pm Post subject:

maybe coz freeBSD isnt evin a linux...
its kernel is totally diffrent...

jcc · Posted: Tue Feb 10, 2004 10:59 pm Post subject:

actually, the fps difference only shows when using this precise quake demo... it's a known phenomenon... Still, no performance boost in any game has been revealed until today... cf the famous FreeBSD NVidia FAQ :)

Trejkaz · Posted: Tue Feb 10, 2004 11:17 pm Post subject:

No, it showed up on glxgears too. ;-)

ChojinDSL · l33t Joined: 07 Jul 2003 Posts: 784

glxgears isnt a very reliable benchmark.

On my old PIII 450mhz with a VoodooIII gfx card I used to get ~1000fps in glxgears. On my new Athlon XP 2600 with a 64mb Geforce 4 mx I only get ~400 fps in glxgears. Makes no difference though, since the games that I play really fly, without a doubt. Glxgears is useless as a benchmark, says nothing about your gaming performance.

blueworm · l33t Joined: 09 May 2003 Posts: 962

I am not sure about glxgears... I did the math.
glxgears is 16% faster in freebsd than in linux.
quake3 is 15% faster in freebsd than in linux.
This Instantly gives glxgears credibility at least as far as quake 3 is concerned.

wolf31o2 · Posted: Wed Feb 11, 2004 7:14 pm Post subject:

The number of active processes and your kernel configuration could play a major part in the difference in framerate.

Do you have pre-empt on in your kernel? Low latency?

What is your AGP status?

Do you have Fast Writes and SBA enabled?
_________________
Ex-Gentoo Developer
Catalyst/Genkernel Development Lead
http://wolf31o2.org

blueworm · l33t Joined: 09 May 2003 Posts: 962

Yes all that stuff is set in linux...
But funny you should mention it because none of that is set in freebsd.
It is a completely stock install.

Lucho[FLCL] · Posted: Thu Feb 12, 2004 6:17 pm Post subject:

How would your "freeBSD Vs. Gentoo Desktop performance" experience be?

'Cause I'm very tempted to install a freeBSD and see it for myself...

Though, your PC is very fast....mine is a PIII 550

jcc · Posted: Thu Feb 12, 2004 6:49 pm Post subject:

guys... we have numbers from glxgears and from a quake demo... Any game numbers?

blueworm · l33t Joined: 09 May 2003 Posts: 962

So far I'm finding it a little unstable...
Its crashed a few times.
Gentoo so far has been rock solid.

jcc · Posted: Fri Feb 13, 2004 5:52 am Post subject:

then let's wait until 5.3 stabilizes, and we'll see if we all have to switch :)

pandaxiongmao · Guru Joined: 29 Sep 2003 Posts: 478 Location: USA

FreeBSD got nice documentation, they almost cover everything, even wireless networking. Sorry it's a bit out of topic.
_________________
CPU: Intel Core 2 Duo
GPU: nVidia GeForce 9800 GT
MB: Asus P5N-E SLI

Rodent · Posted: Sat Feb 14, 2004 5:58 am Post subject:

FreeBSD was on a fresh install which means there was:
no extra processes floating around
no HD fragmentation

not to mention he used different drivers, that test is known to be flawed, glxgears in not accurate (even if it was close in this case, doesn't mean it is reliable; infinite monkeys w/ infinite typewriters can create shakespear but the other stuff is mostly crap); did you use the same DE and WM? Did you have the same hdparms?

Close test shouldn't be supriseing either. Both gentoo and *BSDs compile form source (emerge & ports). In fact the great guy who started gentoo (whats his name) used BSD before starting gentoo and translated alot of the good things about it into a linux enviroment.

And lastly maybe, just maybe the people who did the defaults for FreeBSD know more than you and picked better settings for your computer than you did with gentoo. (this is gentoo's blessing and bain)

BTW, anyone who likes linux but not 'hype' will love FreeBSD

blueworm · l33t Joined: 09 May 2003 Posts: 962

Rodent · Posted: Sat Feb 14, 2004 3:51 pm Post subject:

practicly never fragment I would say is going a little far don't you think? personally I use reiserfs and need to clean it up about once every 2.5 months, but then again I move a ton of bits around. However you are correct sir in the statement that HDs wouldn't have much impact.

Prehaps a much more satisfying reason caim to me; if you had module or a monolithic kernel in linux? If module w/ 3d stuff it could make maybe a 5% difference.

and processes I meant more generally, what was in the linux kernel (there is alot of drviers and things you could bog it down with)

- but none of the things really seem to be the case as it seems you set up the systems correctly. I would like to see some more serious bench marking done on this. of course you would have to do fresh installs with both and use the sane defaults given with only the minium of customization. (I think I might have time and a hd to do it on sunday)

blueworm · l33t Joined: 09 May 2003 Posts: 962

Sir I take my hat of too you.
You are indeed correct, And it gives me no pleasure to assume I am wrong. and that I am truely an asshole.
What you said made me think and I repeated the test using a Gentoo live CD.
Enemy territory. I edited the XF86config so that it had the same settings as my Gentoo install. and my freebsd install.
I again ran glxgears and it give the same results as freebsd.
My conclusion Is that I have screwed up my gentoo and those that know more than me have default setups that are faster than my supposedly tweaked one.
I will take this as a learning experience and hope that I can find my missing 1000 fps.
Thankyou rodent and I apologise for any sarcasm in my previous post...