Benchmark your graphics card (?)

General Forums Off-topic Chat
#1

Hi,
It seems that the new graphics cards have not gained that much power when it comes to pure polygon and texture drawing speed. The new models have better effect processors and speed for them, but the pure strength for great amounts of polygons hasn’t increased in the same ratio. Actually, I haven’t even found any info about how many polygons an nvidia 8800 series card can display. But for a game such as Flight Sim X, most of the last gens best cards were in 1fps range from this gen’s best cards, which isn’t a lot at all (LINK). So, let’s test it with Blender, just for fun.

Here is a blend file. The test is supposed to be ran in solid mode.

Run the benchmark and tell what you get. Top menu: Help > System > Benchmark > Draw.

I got 0.686 s/op 1.46 ops/s 8 iterations

Actually I am not sure what this really measures. But I figure the “draw” should be what the graphics card does. I’d like to see some 8800 series benchmarks.

Anyway, if it’s of any relevance, my system is a Amd64 3500+ with 1.5GB ram and 6600GT 128mb AGP running windows 2000.

#2

Will run some benchmarks on mine when I get home using my 8800gts in Vista, XP and Ubuntu 64. I have run this test on a different blend file and the 8800gts is indeed much faster than the older generation cards.

#3

Well, I have no idea what it means either, but I did it 3 times and got 3 different results, for what its worth.

0.753 s/op 1.33 ops/s 7 iterations
0.629 s/op 1.59 ops/s 8 iterations
0.580 s/op 1.72 ops/s 9 iterations

There is a regular pattern there though, as s/op goes down, iterations and ops/s go up.

core 2 quad @ 2.4 ghz, nvidia 7900gs oc’d to 600mhz, 4gb ram, 64 bit fedora and blender

#4

Did you do it in “solid mode”? Because ZanQdo told me the shaded mode is software rendered, so I changed it to solid mode. Thanks anyway. That’d be a setup I would upgrade to, except that I can’t get a 7900 GS 512mb passive cooled anywhere.

#5

I’m sorry to disapoint you but shaded view is software rendered (the vertex colors are shaded using blender internal) iirc

Also blender’s use of modern OGL faster display mode is almost inexistant, but all this will change with bmesh and GLSL

so… my benchmark (even when its useless information) is this

0.35 s/op, 2.83ops/s, 15 iter

#6

Well, I doubt it’s useless information if you use it in solid view, it does tell something about the speed. Also, you told most of this to me on #blenderchat already but I guess the others deserve to know as well.

#7

BMesh uses vertex arrays wich are 10s of times faster :smiley: I’m waiting for that! and of course GLSL shading wich is fully hardware accelerated

#8

Nope I did it in shaded mode, just left it the way I opened to file. But it looks like there’s no real way to interpret these stats into a meaningful and easy to understand benchmark, unless someoen can explain what each of the terms actually means.

#9

Hey,

I have a 8800GTX and i got the following results:-

0.737455 s/op
1.36 ops/s
7 iterations

My system spec is:-
XP Home
2GB ram
AMD X2 5600+

Hope the info helps :slight_smile:

#10

My supercharge Intel Graphics Controller results (sorry it is not really a “card”):

6.700109 s/op
0.15 ops/s
1 iteration

In about an hour, my lappy will arrive via UPS, and I’ll be able to give you the results of the 8600 GT. :eyebrowlift2:

#11

It really seems there’s no point in the scores.

#12

Intel Core 2 Duo e6400 (2.13 GHz)
4 GB RAM

Two Different Graphics Cards, 8 interations:

nVidia GeForce 7950GT - 512MB

  • 0.683216 s/op; 1.46 op/s
  • 0.658272 s/op; 1.52 op/s
  • 0.653777 s/op; 1.53 op/s

nVidia GeForce 7600GT - 256MB

  • 0.665252 s/op; 1.50 op/s
  • 0.664817 s/op; 1.50 op/s
  • 0.661806 s/op; 1.51 op/s

Hope that helps.

#13

Athlon X2 5200
2 gig Ram
8800GTS 320
Vista 32 bit

0.43 s/op 2.29 ops/s 12 iterations

#14

Look at the scores, is there any logic in them?

#15

I know that with the Geforce 8X00 series of cards, I must open Blender and then Minimize it and maximize it a few times before I benchmark my draw speed. It has something to do with the OpenGL driver for the card. Once I do that, the draw speed improves quite a bit.