your glxgears fps is fast?

[jeonghan]

Hi, everyone

I use SUSE 10.3 linux(586) and lastest nvidia graphic driver(100.14.19)

2 month ago, I tested glxgers. fps over 15000.

CPU : CORE 2 DUO 6300
GPU : 8800 GTX
Mem : 2G
M/B: 965p-ds3 gigabyte .......

Today glxgears points

13473 frames in 5.0 seconds = 2694.510 FPS
13579 frames in 5.0 seconds = 2715.622 FPS
13426 frames in 5.0 seconds = 2685.109 FPS
13644 frames in 5.0 seconds = 2728.653 FPS
13215 frames in 5.0 seconds = 2642.775 FPS

Oooops.. What's the problem?

My machine is same before 2 month ago. Just reinstall from kubuntu to suse.

Could you mind to tell me why glxgears is so low-fps?

[mooninite]

http://wiki.cchtml.com/index.php/Glx...ot_a_Benchmark

[energyman76b]

there are many potential reasons. Working 'cool'n'quit' - or its intel equivalent. Something in the background 'stealing' fps. Some other reasons.

Long story short: glxgears is only usefull if you compare absolutly identical setups where you changed just one thing (like different windowmanager in kde or gnome, or slightly different settings for your scheduler) and even then glxgears is not very usefull. There are a lot better ways to bench your system.

[LubosD]

While glxgears is not a benchmark, this seems to be too low anyway.

Is the direct rendering working? Are you running Compiz or some other composite window manager?

[lloeki]

this seems to be too low anyway

there is no 'too low' or 'too high' value. you can get a ten (or even hundred) scale with the same card on two different computers, or with different cards on the same computer. or roughlt same values with vastly different computers.

e.g I get ~2-3k with a radeon 9000 on both a PIV 2.6HZ and a pII 450mhz.

I have numerous examples of this of numerous cards of numerous brands in various configurations.

my conclusion: glxgears fps value (even the decimal range) means plain sh*t.

how come people just don't get it?

[Skrot]

Originally Posted by lloeki

e.g I get ~2-3k with a radeon 9000 on both a PIV 2.6HZ and a pII 450mhz.

Erm all that your example means is that the glx frame rate is purely limited by the video card. Likely fill rate or there abouts.

While glxgears isn't a benchmark that you can directly say "since I get more FPS than your machine, my computer is better at games than yours is", it is still definitely representative of the throughput of the video card.

On a well-configured system with the specs of the first post, the frame rate should probably be around 20,000. At any rate, no where near as low as 2500 as shown in the first post. I have an 8800GTX with Core 2 Duo 6850, and I get 23000 - 24000. I don't use Compiz or anything like that.

My suggestion is to definitely look into what has changed. Perhaps the only difference in the system is that a 3D compositing window manager is running, slowing down any program like this.

The important question is still: do real games run okay? Try the demo of ETQW perhaps. While the full game runs a bit better for some people, this is a "real" situation where the performance of most parts of the computer will be tested.

[Ralph]

Glxgears is heavily influenced by applications like SuperKaramba (and its terrific themes); on my PC (GeForce 7600GS), SuperKaramba eats 400 FPS!!!

Obviously, I have removed SuperKaramba!

[ThermalGrease]

huhu

not a benchmark but something is wrong with your setup...

e6700, 8800GTX (with lots of stuff running...)

someone@kontoret:~> glxgears
137718 frames in 5.0 seconds = 27543.500 FPS
139002 frames in 5.0 seconds = 27800.400 FPS
139021 frames in 5.0 seconds = 27804.123 FPS
139082 frames in 5.0 seconds = 27816.389 FPS
138014 frames in 5.0 seconds = 27602.674 FPS

someone@kontoret:~>

[energyman76b]

Originally Posted by Skrot

While glxgears isn't a benchmark that you can directly say "since I get more FPS than your machine, my computer is better at games than yours is", it is still definitely representative of the throughput of the video card.

no. That is just wrong. Totally, absolutly wrong.

Please stop it, ok? glxgears says NOTHING about game performance. I had computers who were better at games scoring lower glxgears marks. Or crappy card setups (5200) scoring almost the same as a lot better cards (6600). Change resolution from non-standard 1280x960 to any standard resolution - bang huge fps increase.

glxgears is almost worthless.

http://wiki.cchtml.com/index.php/Glx...ot_a_Benchmark

[Ralph]

Originally Posted by ThermalGrease

huhu

not a benchmark but something is wrong with your setup...

e6700, 8800GTX (with lots of stuff running...)

someone@kontoret:~> glxgears
137718 frames in 5.0 seconds = 27543.500 FPS
139002 frames in 5.0 seconds = 27800.400 FPS
139021 frames in 5.0 seconds = 27804.123 FPS
139082 frames in 5.0 seconds = 27816.389 FPS
138014 frames in 5.0 seconds = 27602.674 FPS

someone@kontoret:~>

ThermalGrease, are you multiplying by 10?

[jeonghan]

Thanks for replying post.

I don't use compositing windows likes compiz or beryl or compiz-fusion.

I understanded that glxgears not benchmark. But I just curious why glxgears fps values not similar between suse and kubuntu.

My wrong setting is cpu scaling. I updated my BIOS and reset yesterday.

Now, It is similar last time.

65214 frames in 5.0 seconds = 13042.798 FPS
65461 frames in 5.0 seconds = 13092.116 FPS
65528 frames in 5.0 seconds = 13105.485 FPS
65529 frames in 5.0 seconds = 13105.764 FPS
65561 frames in 5.0 seconds = 13112.187 FPS
65263 frames in 5.0 seconds = 13052.579 FPS
65488 frames in 5.0 seconds = 13097.569 FPS

thank you

[ThermalGrease]

Originally Posted by Ralph

ThermalGrease, are you multiplying by 10?

nope, is this strange numbers for a 8800GTX and a e6700@3000MGz?
I dont think its so strange... I get a very good score in 3Dmark-06 too...