Nvidia glxgear

[DimGR]

i was wondering what results you nvidia people get from glxgears

this is my ATI 9600 XT with driver 3.12.0


28475 frames in 5.0 seconds = 5695.000 FPS
28453 frames in 5.0 seconds = 5690.600 FPS
28461 frames in 5.0 seconds = 5692.200 FPS
28475 frames in 5.0 seconds = 5695.000 FPS
28483 frames in 5.0 seconds = 5696.600 FPS
28469 frames in 5.0 seconds = 5693.800 FPS
28481 frames in 5.0 seconds = 5696.200 FPS
28469 frames in 5.0 seconds = 5693.800 FPS
28482 frames in 5.0 seconds = 5696.400 FPS
28465 frames in 5.0 seconds = 5693.000 FPS
28475 frames in 5.0 seconds = 5695.000 FPS

[gehidore]

FX5700

90587 frames in 5.0 seconds = 18117.400 FPS
90437 frames in 5.0 seconds = 18087.400 FPS
90531 frames in 5.0 seconds = 18106.200 FPS
90565 frames in 5.0 seconds = 18113.000 FPS
90529 frames in 5.0 seconds = 18105.800 FPS
89872 frames in 5.0 seconds = 17974.400 FPS
90220 frames in 5.0 seconds = 18044.000 FPS
88214 frames in 5.0 seconds = 17642.800 FPS
81082 frames in 5.0 seconds = 16216.400 FPS

;)

[bsm2001]

gehidore FX5700

90587 frames in 5.0 seconds = 18117.400 FPS
90437 frames in 5.0 seconds = 18087.400 FPS
90531 frames in 5.0 seconds = 18106.200 FPS
90565 frames in 5.0 seconds = 18113.000 FPS
90529 frames in 5.0 seconds = 18105.800 FPS
89872 frames in 5.0 seconds = 17974.400 FPS
90220 frames in 5.0 seconds = 18044.000 FPS
88214 frames in 5.0 seconds = 17642.800 FPS
81082 frames in 5.0 seconds = 16216.400 FPS
;)

How small did you make the window?:eek:

bash-2.05b$ glxgears
19763 frames in 5.0 seconds = 3952.600 FPS
19885 frames in 5.0 seconds = 3977.000 FPS
19992 frames in 5.0 seconds = 3998.400 FPS
20017 frames in 5.0 seconds = 4003.400 FPS
20046 frames in 5.0 seconds = 4009.200 FPS
20082 frames in 5.0 seconds = 4016.400 FPS
20052 frames in 5.0 seconds = 4010.400 FPS
20062 frames in 5.0 seconds = 4012.400 FPS

[bwkaz]

Well, since glxgears is NOT a video card benchmark (it's so INSANELY simple that the bottleneck is the CPU by a LONG shot), try running an actual game, or another 3D scene that's actually complicated and textured, instead.

Such as something like SPECviewperf, or Doom 3 on an agreed-upon resolution and detail level, etc. Something more complicated than 3 single-color (lit, but not textured) gears.

[gehidore]

Originally posted by bsm2001
How small did you make the window?:eek:

Ok so I cheated, but bwkaz explains why above

Although I do get 80~150 sometimes 200+ in Ut2k4 and ArmyOps at 1280X1024 with max everything.

[bsm2001]

Doom 3 800x600 I'm getting 15-20 fps full detail.

[timothykaine]

FX 5200 Ultra w/ KDE 3.3, GAIM, Firefox, and a few small apps running.

13227 frames in 5.0 seconds = 2645.400 FPS
14694 frames in 5.0 seconds = 2938.800 FPS
14775 frames in 5.0 seconds = 2955.000 FPS
14904 frames in 5.0 seconds = 2980.800 FPS
15156 frames in 5.0 seconds = 3031.200 FPS
15162 frames in 5.0 seconds = 3032.400 FPS
15160 frames in 5.0 seconds = 3032.000 FPS
15170 frames in 5.0 seconds = 3034.000 FPS
15141 frames in 5.0 seconds = 3028.200 FPS
14893 frames in 5.0 seconds = 2978.600 FPS
15155 frames in 5.0 seconds = 3031.000 FPS

[fatTrav]

Originally posted by bsm2001
Doom 3 800x600 I'm getting 15-20 fps full detail.

Doom3 doesn't count because there isn't a computer on the planet that can really put the game though its paces:p

[DimGR]

Originally posted by bwkaz
Well, since glxgears is NOT a video card benchmark (it's so INSANELY simple that the bottleneck is the CPU by a LONG shot), try running an actual game, or another 3D scene that's actually complicated and textured, instead.

Such as something like SPECviewperf, or Doom 3 on an agreed-upon resolution and detail level, etc. Something more complicated than 3 single-color (lit, but not textured) gears.


can you name some benchmarks for us to get other than just playing the games? :)

[bwkaz]

Like I said there, SPECviewperf would be a decent one.

I'm still trying to get it to work on my machine, though. It requires the C shell by default to run most of its tests, which causes a problem because I don't believe in the C shell. :p

http://www.specbench.org/gpc/downloadindex.html

is the link for the current SPECviewperf package. Don't try to get it from there, though -- the FTP server was giving me about 9 kilobytes/sec. The US mirror was giving about 140 kilobytes/sec, so I used it instead.

[bwkaz]

Never mind. Even after getting the C shell installed (tcsh), none of the tests wanted to work properly for me. Basically they ran their first test dataset thingy, but then they all went into some kind of lots-of-kernel-time loop after doing so, and never came out of it. I don't know why, and at the moment I'm not too sure I want to try to find out either. ;) I suspect my self-compiled glibc or kernel, frankly.

So, I'll revert to Doom 3's "timedemo demo1". On my machine, the first run gave 30.6fps, and the second run gave 33.4fps. The first run was noticeably more jumpy, though -- I'm betting that's because of in-game caching. Maybe it caches texture decompression or something (which I'm sure 768MB of memory doesn't hurt :p).

This was run at 800x600, "high" detail (3rd setting of 4, only "ultra" would be higher), and under advanced options, vsync and antialiasing were off, but everything else was on. I was also using the nvidia-settings program to override the program-defined anisotropic level, setting it to 8x. All other settings were at the installed defaults.

Software: Doom 3 Linux binaries, nVidia 6111 driver, self-compiled kernel 2.6.8. glibc CVS snapshot from Jan. 3, 2004, gcc 3.3.2.

Hardware: Athlon XP 2500+ (Barton core) on an Asus A7V8X-X (Via KT400 chipset). 768MB of DDR333 memory, which claims to be CAS 2 but is actually running at CAS 2.5 (bastards... ;)). Video card is a BFG 6800 GT OC, which I'm not entirely sure if it's actually running overclocked or not. It's supposed to be, but they might do that in the driver. Then again, for Windows, they say to use the official nVidia drivers anyway, so maybe not. NVclock isn't much help, either, since the author posted just yesterday that it doesn't quite work with 6800s yet (though I'm not sure if it would be able to tell whether the current clock speed is the nVidia specified 350MHz, or the BFG 370MHz).

[raz0rblade]

My old Geforce3 Ti500 :)

GLXgears (default size @ 1280x1024)
15742 frames in 5.0 seconds = 3148.400 FPS
20083 frames in 5.0 seconds = 4016.600 FPS
20210 frames in 5.0 seconds = 4042.000 FPS
20215 frames in 5.0 seconds = 4043.000 FPS
20050 frames in 5.0 seconds = 4010.000 FPS
19999 frames in 5.0 seconds = 3999.800 FPS

DOOM3 [timedemo1] at default (low) settings @ 1024x768 window
2148 frames rendered in 148.1 seconds = 14.5 fps

I could have probably gotten a few more FPS from DOOM3, as I was burning a CD when I did it.

[Gnarulf]

I think its some thing wrong with my system, im getting this


3139 frames in 5.0 seconds = 627.800 FPS
3461 frames in 5.0 seconds = 692.200 FPS
3510 frames in 5.0 seconds = 702.000 FPS
3466 frames in 5.0 seconds = 693.200 FPS
3483 frames in 5.0 seconds = 696.600 FPS
3476 frames in 5.0 seconds = 695.200 FPS


I have a FX5200 and im using enlightenment.

[Pafnoutios]

$ glxgears
3631 frames in 5.0 seconds = 726.200 FPS
618 frames in 5.0 seconds = 123.600 FPS
618 frames in 5.0 seconds = 123.600 FPS
618 frames in 5.0 seconds = 123.600 FPS
618 frames in 5.0 seconds = 123.600 FPS
619 frames in 5.0 seconds = 123.800 FPS
618 frames in 5.0 seconds = 123.600 FPS
618 frames in 5.0 seconds = 123.600 FPS
618 frames in 5.0 seconds = 123.600 FPS
618 frames in 5.0 seconds = 123.600 FPS
618 frames in 5.0 seconds = 123.600 FPS
619 frames in 5.0 seconds = 123.800 FPS
618 frames in 5.0 seconds = 123.600 FPS
618 frames in 5.0 seconds = 123.600 FPS
618 frames in 5.0 seconds = 123.600 FPS

I maximized the window as quickly as I could on my 2048x1536 desktop. This is on a gForce4 TI I forget.